This paper estimates the impact of binge drinking on sexual activity among a nationally representative set of high school students during the 1990s and 2000s. The main innovations are explicitly controlling for time-invariant preferences regarding sexual behavior and alcohol use, and eliminating non-drinkers from the comparison group. I find that binge drinking significantly increases participation in sex, promiscuity, and the failure to use birth control, albeit by amounts considerably smaller than implied by merely conditioning on exogenous factors. For all outcomes, impacts rise substantially with binge drinking frequency. Results are similar using alternative comparison groups defined by excluding those who do not exhibit other risky behaviors, and by gender and race/ethnicity, but vary by grade level and over time in different ways for engaging in sex than protective behavior. Effects are much larger for the small fraction of students that has not been taught about
AIDS/HIV infection in school.