Population-level patterns in determinants of risky polysubstance use receive little attention despite rising mortality. We use secondary data from the National Survey on Drug Use and Health, 2002-2019 (N = 706,891), and weighted logistic regression to test trends and associations between gender and marital status on the probability of high-risk polysubstance use (i.e., illicit use of pharmacologically-distinct overdose-associated drugs). Our results show that at the population-level about 0.7% of adults engaged in high-risk illicit polysubstance use in the past month. Our regression analyses reveal that polysubstance use was lowest among married women (pˆ = 0.27%) and highest among never married and previously married men (pˆ = 1.06%). When we predicted polysubstance use from 2002 to 2019, the estimates were remarkably stable for women and married adults, with no statistically significant differences for women across the span and few for married adults. Yet, polysubstance use among men had statistically significant variation (ranging from pˆ = 0.55 to pˆ = 0.96%), and the gender gap was larger and more consistently significant as illicit drugs became prominent later in the study period. This pattern suggests that use among men is shaped by drug contexts and availability. Our findings also highlight the important roles of family structure in shaping high-risk drug use. Rising polysubstance use mortality may not stem from rising rates of use or large changes in patterns among high-risk groups, but may be related to the growing share of non-married adults, especially men, who are high-risk relative to married adults and women.