Despite the common use of Clover traps to capture white-tailed deer (Odocoileus virginianus), little published information exists quantifying trap success, trap selectivity (sex-age selection), or weather correlates of trap success. We quantified these relationships using white-tailed deer data from 3 study sites in the Lower Peninsula of Michigan, USA, during winters 2001–2007. We captured 610 deer in 8,569 trap-nights; pooled capture success was 0.07 deer/trap-night, although site-year success varied from 0.021 to 0.086. We compared sex-age classes (fawn [pooled by sex], ad M, ad F) captured with sex-age classes estimated to be available on each study site during each capture season. We used generalized linear mixed modeling to construct 19 a priori models to describe probability of capture success as a function of weather covariates (daily min. temp [° C], daily snowfall [cm], daily snow depth [cm]), Julian day, site, and year. General selection patterns included fawns captured more than expected and adult males captured less than expected; adult females were generally neutrally selected. The quasi-Akaike's Information Criterion best model within our set was described by the global model without Julian day and contained all 3 weather covariates and site-year effects. Our model provided some evidence that as daily snow depth increased, probability of capture increased; the positive effect of daily snowfall on capture probability was dependent on decreasing daily minimum temperature. Our results may be used to increase efficacy of deer capture programs by researchers and managers through informed decision-making about when to allocate effort (e.g., if extreme winter weather conditions are predicted) and when to consider alternative methods (e.g., if capture of ad M is an objective).