Analog amplitude-modulated Holter devices are in widespread use for arrhythmia detection, but their reliability remains questioned for ST-segment analysis. In contrast, recently developed digital Holter devices immediately digitize and analyze the electrocardiogram (ECG) on-line and may therefore be more reliable for ST-segment analysis. To test this hypothesis, the results of digital, on-line, 2-channel ST-segment analysis were directly compared to those of analog amplitude-modulated recordings in identical leads (CM5 and CM3), using a stripchart recorder meeting the American Heart Association specifications as the standard. Thirty-five patients (25 with coronary artery disease and 10 control subjects) underwent graded treadmill exercise testing. The reference ECG mean value for ST-segment depression in CM5 was -1.4 +/- 1.2 mm and in CM3 -0.5 +/- 1.2 mm. For digital analysis, the mean values and correlation coefficients for CM5 were -1.5 +/- 1.1 mm (r = 0.97) and for CM3 -0.8 +/- 1.3 mm (r = 0.93). For analog recording, the results for CM5 were -2.1 +/- 1.7 mm (r = 0.88) and for CM3 -1.3 +/- 1.9 mm (r = 0.85). The mean duration of ST-segment depression with the reference ECG was 7.1 +/- 4.1 minutes. Digital Holter showed a significantly better agreement (7.4 +/- 4.4 min, r = 0.97) than analog Holter (9.6 +/- 5.6 min, r = 0.84). Because analog amplitude-modulated Holter recordings overestimated the degree and duration of ischemic episode, digital, on-line and full disclosure devices should be preferred to assess myocardial ischemia.