The predictive power of home blood pressure (BP) in the evening compared with home BP in the morning and office BP has been controversial. The predictive power of evening BP was compared to that of morning BP and office BP. The likelihood ratio test between one model containing a single BP index with traditional risk factors and a similar model further containing another BP index was used to assess whether the additional BP index significantly improved the adequacy of the model. Of 3266 patients with mild-to-moderate hypertension who were on antihypertensive medications (men 50.6%, age 59.5 ± 10.0 years), 58 experienced a major adverse cardiovascular event during a median follow-up of 7.1 years. The hazard ratios for a one standard deviation increment of evening home systolic/diastolic BP were 1.26 (0.98-1.62)/1.43 (1.09-1.88) in the baseline untreated period and 1.46 (1.17-1.81)/1.63 (1.26-2.11) during the on-treatment follow-up period. When evening BP at baseline and that during follow-up were included in the same model, only the latter significantly improved the prediction models (P = 0.006/0.005 for systolic/diastolic BP). Then, evening home BP vs. morning BP during follow-up was tested. The former did not improve the prediction models (P > 0.2), but the latter significantly improved the models (P ≤ 0.048). Similarly, when evening home BP and office BP during follow-up were analyzed, only the former significantly improved the prediction models (P ≤ 0.015). In conclusion, evening BP could be a more potent predictor than office BP, but it was inferior compared to morning BP in the treatment of mild-to-moderate hypertensive patients.