Objectives: To examine the usefulness of time domain heart rate variability (HRV) measurements by a simple graphical method, the differential index (DI), in prognostic assessments of patients with chronic stable angina pectoris. Methods: HRV measurements in the time domain by DI were compared to conventional measurements of standard deviation of all normal-to-normal intervals (SDNN), percent of differences between adjacent normal RR intervals >50 ms (PNN50) and square root of the mean of the sum of squares of differences between adjacent normal RR intervals (RMSSD) from 24-hour ambulatory electrocardiographic recordings in 678 patients in the Angina Prognosis Study in Stockholm. The patients received double-blind treatment with metoprolol or verapamil. Main outcome measures were cardiovascular death or non-fatal myocardial infarction during follow-up (median 40 months). Results: Patients suffering cardiovascular death (n = 30) had lower DI, SDNN and PNN50 (all p < 0.001). In a multivariate Cox model, DI below median independently predicted cardiovascular death (p = 0.002), as did SDNN (p = 0.016) and PNN50 (p = 0.030), but not RMSSD (p = 0.10). The separation of survival curves was most pronounced and specificity was slightly better with DI. DI and PNN50 increased with metoprolol but not verapamil treatment. Short-term treatment effects were not related to prognosis. Conclusions: Low time domain HRV carries independent prognostic information regarding cardiovascular death in stable angina pectoris. The simple DI method provided equally good or better prognostic information than conventional, more laborious HRV methods.