Introduction: Major gaps exist in the measurement of cannabis exposure. The accuracy of self-reported cannabis and tobacco dose per joint is poorly characterized and has never been investigated following acute cannabis/tobacco exposure. Using an innovative "Roll a Joint" paradigm, this study aims to (1) compare estimated and actual dose of cannabis and tobacco per joint at baseline and (2) examine the acute effects of cannabis and/or tobacco on estimated and actual dose. Materials and Methods: We investigated this by using a randomized, double-blind, placebo-controlled crossover 2 (active cannabis, placebo cannabis)×2 (active tobacco, placebo tobacco) design in a laboratory setting. Participants were 24 recreational cousers of cannabis and tobacco. At baseline, they were asked to measure out the amount of cannabis and tobacco they would put in an average joint for themselves (dose per joint). Then, on each of four drug administration sessions, participants were again asked to do this for a joint they would want to smoke "right now." Self-reported and actual amount was recorded (g). Results: At baseline, the estimated amount of cannabis per joint (0.28±0.23 g) was double the actual amount (0.14±0.12 g) (p=0.003, d=0.723). No difference emerged between estimated (0.43±0.25 g) and actual (0.35±0.15 g) (p=0.125) amount of tobacco per joint. Compared to placebo, active cannabis reduced the actual dose of both cannabis (p=0.035) and tobacco (p<0.001) they put in a joint. Participants accurately estimated this reduction for tobacco (p=0.014), but not for cannabis (p=0.680). Conclusions: Self-reported dose per joint is accurate for tobacco but dramatically overestimates cannabis exposure and therefore should be viewed with caution. Cannabis administration reduced the amount of cannabis and tobacco added to joints, suggesting a reduction in dose during a smoking session. The "Roll A Joint" paradigm should be implemented for better accuracy in assessing dose per joint.