The aims of this study were to identify predictors of health practices of pregnant women in the literature reviewed, to use meta-analysis to ascertain the mean effect size (ES) across studies between each predictor and health practices, and to examine four moderators on each predictor–health practices relationship. Using preferred reporting items for systematic reviews and meta-analyses guidelines for the literature assessed, 32 published studies or doctoral dissertations completed between 1992 and 2015 met the inclusion criteria. Twelve predictors were identified, and each predictor in relation to health practices was subjected to meta-analysis. One predictor (maternal–fetal attachment) of health practices had a large ES, two predictors (depression and stress) had medium ESs, six predictors (income, education, parity, social support, employment, and age) had small ESs, and three predictors (gestational age, marital status, and race) had trivial ESs. Findings are interpreted relative to health practices in pregnant women.