The U.S. cannabis landscape has changed dramatically (e.g., increasing state legalization; products with high delta-9-tetrahydrocannabinol (THC) potency (Borodovsky et al., 2024; Budney et al., 2024; Chen et al., 2025; Cooper & Haney, 2009; ElSohly et al., 2024; Hall et al., 2023)), while rising rates of adult cannabis use and cannabis use disorder (CUD) (Compton et al., 2024, Hasin et al., 2016, Substance Abuse and Mental Health Services Administration, 2024) have outpaced our knowledge of the public health effects of cannabis use (Lake et al., 2025, National Institute on Drug Abuse, 2025), leaving an urgent need for accurate information about the relationship between THC exposure levels to health harms and benefits. To better understand this relationship, large samples of cannabis consumers must be studied. Online surveys offer a rapid, feasible way to ascertain the necessary information, but they require reliable, valid self-report measures of cannabis exposure, i.e., mgTHC. However, the widening variety of cannabis products without standardized labels and increasingly complex use patterns complicate measurement of mgTHC exposure. Earlier national surveys assessed frequency (i.e., days used) but not quantity (mgTHC). Prior efforts to measure mgTHC using visual reference images presented online had confusing items (Walsh et al., 2023) and psychometric testing that was either limited to student samples (Cuttler and Spradlin, 2017, Gette et al., 2023, Sullivan et al., 2021), or not reported at all (Goodman et al., 2019, Hammond et al., 2022), leaving an unmet need for a self-report measure with demonstrated reliability and validity for use in large-scale online surveys.

To begin to address this need, we have used a rigorous, mixed-methods approach to develop the Cannabis Exposure Inventory (CEI). The CEI is a self-report instrument that captures heterogeneous consumption patterns using detailed items and multiple reference images to assess amounts and potencies (%THC) of consumption of 7 categories of products and routes of administration, e.g., Smoking Flower; Dabbing Concentrates; Edibles. A computer algorithm combines these detailed items to produce a standardized measure of mgTHC per using day. Earlier stages in CEI development and its mgTHC measure included cognitive interviews to ensure that participants understood the survey items (Walsh et al., 2023) and validity testing, including determination of the impact of different quantity reporting units, e.g., grams versus hits (Borodovsky et al., 2023, Borodovsky et al., 2024), and the relationship between mgTHC per using day and CUD severity among daily consumers (Borodovsky et al., 2025). For example, in a prior sample of daily cannabis consumers assessed with the CEI (n=4,134), greater daily mgTHC predicted significantly higher CUD criteria count and higher odds of mild, moderate or severe CUD (Borodovsky et al., 2025).

A critical phase in the development of self-report measures is assessment of test-retest reliability, i.e., the stability of repeated within-individual responses over a short period of time. This is an important indication of whether an instrument produces replicable measurements. We therefore conducted a test-retest study of the CEI mgTHC measure and key additional items (i.e., methods of use, age at onset of cannabis use). We conducted this study in an online sample of 511 adult cannabis users. In addition to determining reliability in the full sample, we explored reliability within subgroups as defined by sex, age, race/ethnicity, legal status of cannabis in participants’ state of residence, and self-defined reason for cannabis use (medical-only, recreational-only, and medical and recreational).