The Berlin Questionnaire and Epworth Sleepiness Scale (ESS) are commonly used to screen for sleep apnea in non-pregnant populations. We sought to evaluate the Berlin and ESS in pregnancy and to determine whether an alternative screening approach could better detect sleep apnea in pregnant women. Pregnant women at high risk for sleep apnea (women with chronic hypertension, pre-gestational diabetes, obesity, and/or a prior history of preeclampsia) completed a sleep survey composed of the Berlin and ESS, and participated in an overnight sleep evaluation with the Watch-PAT100 (WP100), a wrist-mounted device designed to diagnose sleep apnea, defined as an apnea hypopnea index ≥ 5. Using multivariable statistics, demographic, clinical, and subjective symptoms that were independently associated with sleep apnea were determined and a prediction rule for the presence of sleep apnea was developed. The predictive capacity of this newly developed system was compared to that of the Berlin and ESS using receiver-operating curve (ROC) statistics. Of the 114 women who participated and had a valid WP100 study, 100 completed the Berlin and 96 the ESS. The Berlin and ESS did not accurately predict sleep apnea in this high-risk pregnancy cohort, with ROC area under the curves (AUC) of 0.54 (p = 0.6) and 0.57 (p = 0.3), respectively. Conversely, a model incorporating frequent snoring, chronic hypertension, age, and body mass index performed significantly better (AUC 0.86, p > 0.001). The Berlin and ESS are not appropriate tools to screen for sleep apnea in high-risk pregnant women. Conversely, our four-variable model more accurately predicts sleep apnea in pregnancy.