This study determined the reproducibility of cardiovascular, respiratory, and metabolic responses to submaximal cycle ergometer exercise at two power outputs (50 W and 60% VO2max) on each of two separate days in a sample of 390 subjects (198 men and 192 women) participating in the HERITAGE Family Study. The same protocol was conducted across 3 d in an intracenter quality control substudy which included an additional 55 subjects. Reproducibility estimates included technical error, coefficient of variation, and intraclass correlation for each of the selected variables for both subject populations. Further, since the data were collected across four clinical centers, intraclass correlations were also computed separately by clinical center. Most variables were highly reproducible, with coefficients of variation below 9% and intraclass correlations over 0.80. These results were consistent for both subject populations and across all four clinical centers. Reproducibility indicators were generally better at the higher power output. It is concluded that within-subject day-to-day variation and measurement unreliabilities are generally small compared with the between-subject variance in the response to submaximal exercise at each of the clinical centers of the HERITAGE Family Study.