RESUMEN
To analyze whether heart rate variability is reproducible after maximal exercise, 11 men (22.1±3.2 years) performed four incremental exercise tests followed by passive or active recovery. There was high reliability (intraclass coefficient correlation: 0.72-0.96) and fair-to-excellent agreement (coefficient of variation: 7.81-22.09%) in passive recovery, as well as moderate-to-high reliability (intraclass coefficient correlation: 0.50-0.87) and good agreement (coefficient of variation: 11.08-20.89%) in active recovery for LnRMSSD index. There was moderate-to-high reliability (intraclass coefficient correlation: 0.51-0.81) and good agreement (coefficient of variation: 10.41-18.87%) in most of the analyzed time points, in both recovery types for LnSDNN. In both types of recovery, the time domain heart rate variability 5-10 min indices (passive: intraclass coefficient correlation : 0.87-0.88; coefficient of variation: 7.67-13.44%; active: intraclass coefficient correlation 0.59-0.80; coefficient of variation: 14.62-16.26%) presented higher intraclass coefficient correlation and lower coefficient of variation than the spectral heart rate variability indices (passive: intraclass coefficient correlation: 0.71-0.87; coefficient of variation: 12.33-34.21%; active: intraclass coefficient correlation: 0.46-0.77; coefficient of variation: 24.41-105.12%). The LnRMSSD and LnSDNN indices analyzed in 30 s segments and the heart rate variability 5-10 min indices after maximal exercise in untrained healthy men showed satisfactory reproducibility, regardless of the type of recovery, with the time-domain indices showing higher reproducibility than the frequency-domain indices.