Official figures show that life expectancy in Costa Rica is longer than in the United States (US), in spite of the fact that per capita health expenditure is only one-tenth that of the US. To check whether this is for real and to explore some of its determinants, 900 Costa Ricans aged 60+ were followed from 1984 to 2001. Follow-up household visits were made, deaths were tracked in the national death registry, and survival status in the voting registry was double-checked. In addition, the survivors were contacted in 2002. Two-thirds of the panel had died by December 2001. Kaplan-Meier curves, life tables and Cox regression were used to analyse the panel's survival. Mortality in the panel was slightly higher than the Costa Rican average and similar to that in the US, confirming the exceptional longevity of Costa Ricans. Survival was substantially lower among unmarried men and individuals with limited autonomy at the beginning of the study. The effect of socioeconomic status is weak. Insurance effects seem to be confounded by selection biases.