Cortisol is used in research as a biomarker of psychological stress. Logistical considerations argue for collecting as few samples as possible, balanced against diurnal rhythms and intra-individual variations. 100 pregnant women gave five saliva samples a day for 3 days, at waking, 30 min after waking, and 11:00 a.m., 5:00 p.m., and 9:00 p.m. Timing of collection was confirmed by monitors. Another sample was taken during a clinic visit. Using the 15 measures as the gold standard, correlations and mean area under the curve (AUC) were compared with subsets and the single clinic sample to evaluate alternate collection protocols. Five samples in 1 day, or protocols involving morning and night samples, had the highest correlations with mean AUC (correlation coefficient ranging from 0.82 to 0.88). Standardizing the clinic measurement to a single time of day did not substantially improve correlations with mean AUC. Correlations with measures of reported stress were also not strong.