Measuring health complaints by administrating a single report is common. Our aim was to assess variation in pain and other subjective complaints over an extended period, whether a single-sample produces representative data, and determine associations between complaints. Health-complaint reports were collected from postal workers at monthly intervals over a period of 32-34 consecutive months (1997-2000). We computed six compound complaint-severity indices of 30 complaint-severity scores (intensity score x duration score, scale 0-9). In 67% of the scores, the complaints exhibited larger deviation from a reference (12 consecutive reports in the last 24 months of the study period) when using one report from the respective reference period compared with the mean of two consecutive reports. Four consecutive samples were needed to obtain agreement for 95% of the data when the criterion of accepted deviation from the reference was set to +/-1.0. Neither inspection of graphs nor statistical tests revealed any seasonal pattern or trend on either a group or individual level. The musculoskeletal and psychological complaint-severity indices correlated strongly (rs > 0.66). Correlations between the different somatic indices were generally weak or moderate (rs < 0.55). The initial report produced higher complaint ratings than subsequent reports did. Due to large intra-individual complaint variability and higher complaint-severity level exhibited on the initial report compared to those that followed, measuring subjective health with a single-sample approach does not produce data representativeness for average complaints over a period. More than two samples should be collected when the purpose is to reveal changes in health.