Background: We posit that as rates of hypertension (HT) control improve nationally, recidivism from controlled to uncontrolled HT will emerge as a major obstacle to ongoing improvement in overall population rates of HT control. To probe this hypothesis, we examined rates and predictors of HT recidivism in adults with baseline adequate HT control who are receiving care at medical groups that have overall high rates of HT control. Methods: Study subjects were adults with controlled hypertension at baseline based on two consecutive visits with normal BP readings (<140/90 mmHg or <130/80 mmHg for those with diabetes) at Medical Group A (MG-A; N=12,766) and Medical Group B (MG-B; N=9,768). We classified HT recidivism after follow-up for 4-16 months after the initial BP measures using the mean of the last two BP readings for each patient. Results: At baseline, the proportion of adults with HT who were at BP goals was 55% at MG-A, and 66% at MG-B. HT recidivism occurred in 19% of subjects with baseline controlled HT (based on two consecutive baseline visits) at MG-A and 13% of subjects at MG-B. At MG-A, men (p=.008) and those with higher BMI (p<.001) were more likely to have HT recidivism. At MG-B, those who were younger (p<.001) and with higher BMI (p<.001) were more likely to have BP recidivism. At both MGs, DBP was more likely to rise to uncontrolled levels in those age 50 or under compared to older age groups, while SBP was more likely to rise to uncontrolled levels in those age 65 or older compared to younger age groups. Conclusions: HT recidivism occurred in 13% to 19% of patients with previously controlled HT over a mean of 14 months of follow-up time. In medical groups with relatively high levels of baseline HT control, HT recidivism represents a brake on efforts to improve overall BP control. Effective strategies to minimize HT recidivism have great potential to improve overall levels of HT control on a population basis, and could improve HT-related quality measures at the medical group and individual provider level.