The metric properties of health-related quality of life measures are typically evaluated on selected samples and assumed to hold across different population groups. We assessed the extent to which the measurement properties of the Spanish version of the Nottingham Health Profile (NHP) were stable across sociodemographic, clinical and geographical characteristics. We collected information from all available studies using the NHP in Spain (1986-1995), and obtained data from 9419 individuals. We examined data completeness and distribution, as well as reliability and construct validity. The percentage of missing dimensions was lower than 5%, but increased with age and poor health status. Large ceiling effects in scores were observed for social isolation and energy dimensions, being largest for younger ages and individuals reporting "very good/good" health. Reliability was higher than 0.7 in all population groups considered, except for social isolation and energy dimensions in some subgroups. Mean NHP scores correlated with self-rated overall health (r = 0.48), but they varied substantially by age among those rating their health as "very good/good." We conclude that NHP is adequate for all Spanish populations. Nevertheless, some caution is needed because the reliability of dimension scores is not sufficient for individual purposes. And also, because a non-negligible high ceiling effect renders the instrument inefficient to measure changes over time among healthy populations.