BACKGROUND We tested the effect of initiating marijuana and alcohol use at varying frequencies on academic indices. METHODS In a sample of 26,475 grade 9‐12 students with at least 2 years of linked longitudinal data from year 1 (Y1: 2012‐2013), year 2 (Y2: 2013‐2014), and year 3 (Y3: 2014‐2015) of the COMPASS study, separate multinomial generalized estimating equations models tested the likelihood of responses to measures of academic goals, engagement, preparedness, and performance when shifting from never using alcohol or marijuana at baseline to using them at varying frequencies at follow ‐up. RESULTS Students who began using alcohol or marijuana were less likely to attend class regularly, complete their homework, achieve high marks, and value good grades, relative to their abstaining peers. Changing from abstaining to rare/sporadic‐to‐weekly drinking or rare/sporadic marijuana use predicted aspirations to continue to all levels of higher education, and initiating weekly marijuana use increased the likelihood of college ambitions, while more regular marijuana use reduced the likelihood of wanting to pursue graduate/professional degrees, over high school. CONCLUSIONS The importance of delaying or preventing substance use is evident in associations with student performance and engagement. The influence on academic goals varied by substance and frequency of initiated use.