The aim of this systematic review with meta-analysis was to analyse heavy strength training effects on physiological determinants (i.e., VO2max, pVO2max, MMSS, cycling efficiency, anaerobic capacity, and anaerobic power) of endurance cycling performance (i.e., time to exhaustion and time trial). Heavy strength training produced a small effect on cycling efficiency and a moderate effect on anaerobic power and cycling performance. Heavy strength training may improve cyclist performance, with underlying mechanisms involving improved cycling efficiency and anaerobic power. The results discussed in the following paragraphs are in line with the model proposed by Joyner and Coyle (Joyner and Coyle 2008), who propose that the interaction of VO2max, MMSS, non-oxidative energy contribution (i.e., anaerobic power and anaerobic capacity) and cycling efficiency determines endurance performance.

Heavy strength training did not improve VO2max (ES = − 0.041, p = 0.773). The Fick equation indicates that VO2max depend on maximal cardiac output (Qmax) and the arteriovenous oxygen difference (a-vO2 difference), thus VO₂max = Qmax × a-vO2 difference (Skattebo et al. 2020a). The a-vO2 difference, dependent on arterial oxygen content (CaO2) and mixed venous oxygen content (CvO2), is similar in trained and untrained participants (Ekblom et al. 1968), suggesting that training may have a limited impact on VO2max through adaptations in the a-vO2 difference (Skattebo et al. 2020a). However, improved skeletal muscle fibres capillary and mitochondrial density may increase oxygen extraction (i.e., CvO2), thus VO2max, and this seems a particularly relevant mechanisms in highly trained individuals (Skattebo et al. 2020a). Skeletal muscle fibres capillary and mitochondrial density improvement is dependent on exercise intensity, requiring mainly efforts that challenge the muscle oxidative capacity (Skattebo et al. 2020b), such as endurance training efforts close to VO2max. In contrast, 16 weeks of heavy strength training in highly trained cyclists did not change capillarisation (Aagaard et al. 2011). Moreover, it has been proposed that strength training, when combined with endurance training, may impair mitochondrial remodelling (Zhao and Gao 2024). However, this interference effect appears to be limited in trained individuals and may even be beneficial in untrained individual (Zhao and Gao 2024). Overall, the primary mechanism underlying improvements in VO2max after training seems to be the Qmax (i.e., product of stroke volume and maximal heart rate) (Montero et al. 2015; Montero and Díaz-Cañestro 2016). Maximal heart rate changes with training does not seem to be related to changes in VO2max (Skattebo et al. 2020a), and increases in heart rate may not be energetically favourable for myocardial performance (Heinonen 2025). Therefore, increased stroke volume seems a key mechanism for high VO2max (Skattebo et al. 2020a). Stroke volume improvement may be achieved through endurance training efforts requiring (near to) maximal stroke volume (Heinonen 2025), which ranges from ~ 40 to 100% VO2max in untrained to trained individuals, respectively (Vella and Robergs 2005). Therefore, heavy strength training may fail to appropriately stimulate the main underlying mechanisms associated with VO2max (Heinonen 2025), in line with our meta-analyses.

The MMSS was not affected by heavy strength training (ES = 0.069, p = 0.308). This physiological determinant is defined as the highest oxidative metabolic rate that can be maintained during continuous exercise (Jones et al. 2019) and is independently affected by convective oxygen supply, diffusive oxygen transport and oxygen utilisation (Goulding and Marwood 2023). Convective oxygen supply refers to the transport of oxygen through the circulatory system to active muscles (Goulding and Marwood 2023). This process is influenced by the duty cycle of muscle contraction, as blood flow is restricted during contraction and increases during relaxation due to changes in intramuscular pressure and compression of blood vessels (Goulding and Marwood 2023). This mechanism may explain the transient increase in muscle blood flow observed after the pedal thrust phase of the crank cycle, probably because of a brief occlusion during contraction (Takaishi et al. 2002). Supporting this idea, a study found a significant pre-post intervention effect on MMSS (measured as power output at 4 mmol BLa−1), which correlated with a shift towards early peak torque during pedal stroke (r = − 0.50, p = 0.05) (Rønnestad et al. 2015), suggesting a reduced duration of blood flow occlusion and, consequently, enhanced convective oxygen supply. In contrast, several studies did not report a significant effect of heavy strength training on MMSS (Bishop et al. 1999; Hausswirth et al. 2010; Sunde et al. 2010; Aagaard et al. 2011; Beattie et al. 2017; Ji et al. 2022). This discrepancy may be explained by the other two limiting factors of MMSS (i.e., diffusive oxygen transport and oxygen utilization). Diffusive oxygen transport refers to the movement of oxygen from the capillaries to the muscle mitochondria and is primarily determined by muscle capillarity (Mitchell et al. 2018; Goulding and Marwood 2023), particularly the capillarization of type I muscle fibres (Mitchell et al. 2018). As previously discussed, no significant changes in capillarization have been observed following heavy strength training in highly trained cyclists (Aagaard et al. 2011), which may explain the lack of improvement in this component of oxygen delivery. As for oxygen utilisation, this is determined in part by the oxidative capacity of the muscle (Jones et al. 2019; Goulding and Marwood 2023; Peden et al. 2024), with MMSS showing a strong correlation to mitochondrial content (r = 0.88, p < 0.001) (Peden et al. 2024). However, two studies (Bishop et al. 1999; Vikmoen et al. 2016) failed to find significant expression of aerobic enzymes (specifically citrate synthase, hydroxyacyl-CoA dehydrogenase, cytochrome c oxidase subunit IV, and 2-oxoglutarate dehydrogenase). Considering the above, improvements in convective oxygen delivery (potentially due to increased blood flow from reduced muscle contraction time during the pedal thrust phase), appear less critical for enhancing MMSS than improvements in diffusive oxygen transport and, primarily, oxygen utilization capacity. Therefore, it is possible that heavy strength training is of limited relevance for improving MMSS.

On the other hand, “interference effect” of concurrent training, arising from potential conflicts between molecular regulators of muscle metabolism and acute residual fatigue from strength training) may impair aerobic adaptations (Coffey and Hawley 2017). Hickson’s seminal study (Hickson 1980), observed a plateau in strength gains (1RM) without changes in VO2max, but a separate study the same year (Hickson et al. 1980), reported a modest VO2max increase (~ 4%) and a substantial improvement in time to exhaustion at VO2max (~ 47%). These results suggest that while interference may hinder hypertrophy and strength, it does not necessarily impair oxidative metabolism (Coffey and Hawley 2017). This is supported by meta-analyses showing no significant effects of heavy strength training on VO2max and MMSS in endurance runners and skiers (Castañeda-Babarro et al. 2022; Llanos-Lagos et al. 2024a)., evidence indicates that heavy strength training does not meaningfully improve VO2max or MMSS but also does not negatively affect these factors. Therefore, the exclusion of heavy strength training should not be based solely on concerns about interference with aerobic adaptations.

The pVO2max, a metric that reflects the interaction between VO2max, cycling efficiency and anaerobic performance (Jones and Carter 2000), it is considered a good predictor of cycling performance (Balmer et al. 2000; Bentley et al. 2001). In our analysis, we did not observe a significant improvement in this metric (ES = 0.164, p = 0.308), despite finding improvements in cycling economy and anaerobic power. It is possible that a significant effect on pVO2max was not found because an improvement in cycling efficiency and/or anaerobic power is not a sufficient stimulus to increase this variable, as an increase in VO2max would be. On the other hand, the different protocols used could affect the accuracy of the detection of the changes produced by heavy strength training. For example, there were studies that measured pVO2max with long (i.e., ≥ 3 min) (Jackson et al. 2007; Levin et al. 2009; Ji et al. 2022) and short (i.e., ≤ 1 min) (Rønnestad et al. 2010a, b, 2015; Hausswirth et al. 2010; Vikmoen et al. 2016; Beattie et al. 2017; Del Vecchio et al. 2019) duration stages. In fact, it is proposed that short-duration stages are more suitable for measuring pVO2max (Panissa et al. 2017). However, researchers (Sunde et al. 2010) who also measured time to exhaustion at pVO2max intensity found a significant improvement after the inclusion of heavy strength training. It is, therefore, suggested that future research should consider using appropriate protocols to determine the changes in pVO2max generated by strength training.

Our meta-analysis showed that heavy strength training improves cycling efficiency (ES = 0.353, p = 0.012). For an explanation of these results, it is necessary to discuss the possible mechanisms by which this training method could improve cycling efficiency. From strength training, we can achieve different neuromuscular adaptations, which we can classify (although not independent of each other) as neurological adaptations (i.e., central adaptations) and morphological adaptations (i.e., peripheral adaptations) (Folland and Williams 2007; Suchomel et al. 2018). Within the neurological adaptations, it is known that heavy strength training can improve motor unit recruitment and firing frequency, which may result in an improvement in maximal muscle strength (e.g., maximal voluntary contraction) and rate of force development (RFD) (Aagaard et al. 2002, 2011). About maximal muscle strength, several studies included in the analysis have reported an improvement in maximal muscle strength in lower limbs, measured as maximal force in isometric half squat (Rønnestad et al. 2010b, 2015), isometric quadriceps (Aagaard et al. 2011), isometric mid-thigh pull (Beattie et al. 2017), isometric leg press, leg extension and leg curl (Ji et al. 2022). Therefore, if this gain in maximal muscle strength (which was measured off the bike) were to be translated on the bike, the force required for each pedal thrust would become a smaller percentage of the new maximal force (Hickson et al. 1988; Aagaard et al. 2011). Along the same lines, in accordance with the principle of motor unit recruitment (Henneman et al. 1965) and considering that slow fibres (i.e., fibres I type) are more efficient than fast fibres (i.e., type II fibres) (Coyle et al. 1992), a lower relative force demand would favour a greater involvement of slower fibres. This would delay the activation of the faster fibres, thus improving cycling efficiency. Regarding RFD, two studies reported an improvement in this measure during squat 90° (Sunde et al. 2010) and isometric quadriceps (Aagaard et al. 2011). This improvement could be related to an earlier peak torque improvement during the propulsive phase of pedalling (Aagaard et al. 2011), as has been found in a later study (Rønnestad et al. 2015). For example, it has been reported that in isometric contractions the energy cost is higher at the beginning of the muscle contraction than during the maintenance phase of contraction (Russ et al. 2002). Therefore, if this metabolic behaviour occurs during muscle contraction in the pedal stroke, an early torque peak would reflect a decrease in force generation time, increasing force maintenance time, decreasing the energy cost of muscle contraction and thus improving cycling efficiency. Moreover, Hansen et al. (2012) (a follow-up to a study in this meta-analysis (Rønnestad et al. 2011)) found that heavy strength training reduced negative crank torque during the upstroke phase. This was linked to increased hip flexor activation, possibly due to improved RFD, tendon stiffness, or muscle activation timing. Similarly, professional cyclists exhibit lower negative crank torque in the upstroke than elite and club-level cyclists (García-López et al. 2016). Enhanced hip flexor activation may lessen extensor muscle workload (Hansen et al. 2012), potentially improving cycling efficiency. However, these studies did not directly assess cycling efficiency or its mechanisms, highlighting the need for further research.

Morphological changes such as a change in fibres IIx to IIa have also been postulated to improve cycling efficiency (Aagaard and Andersen 2010; Rønnestad and Mujika 2014; Mujika et al. 2016). For example, two studies reported an increase in type IIa fibres with a decrease in type IIx fibres after 16 weeks (Aagaard et al. 2011) and 11 weeks (Vikmoen et al. 2016) of heavy strength training in male and female cyclists, respectively. While another study showed no changes in muscle fibres characteristics (i.e., changes in fibre percentage, fibre area and fibre least diameter) (Bishop et al. 1999). The increase in type IIa fibres at the expense of IIx fibres, which are less prone to fatigue and have a higher power output compared to slower fibres (Bottinelli et al. 1999), could improve cycling performance (Aagaard et al. 2011; Vikmoen et al. 2016) and cycling efficiency (Vikmoen et al. 2016). However, these changes have not been accompanied by an improvement in cycling efficiency (Aagaard et al. 2011) or have not correlated with the change in cycling efficiency (Vikmoen et al. 2016). Therefore, further research is required to clarify the effect on fibre type change and its relationship to cycling efficiency. In addition, an increase in the CSA of the quadriceps femoris muscle has been correlated with cycling efficiency (r = 0.535