Scores on an optimistic-pessimistic personality scale have been associated with mortality, but optimism and pessimism scores are separable traits and it is unclear which has effects on health or longevity. The Life Orientation Test (LOT), containing items for optimism and pessimism, was included in a twin study on health of Australians aged over 50 in 1993–1995. After a mean of 20 years, participants were matched against death information from the Australian National Death Index. 1,068 out of 2,978 participants with useable LOT scores had died. Survival analysis tested for associations between separate optimism and pessimism scores and mortality from any cause, and from cancers, cardiovascular diseases or other known causes. Age-adjusted scores on the pessimism scale were associated with all-cause and cardiovascular mortality (Hazard Ratios per 1 standard deviation unit, 95% confidence intervals and p-values 1.134, 1.065–1.207, 8.85 × 10^–5 and 1.196, 1.045–1.368, 0.0093, respectively) but not with cancer deaths. Optimism scores, which were only weakly correlated with pessimism scores (age-adjusted rank correlation = − 0.176), did not show significant associations with overall or cause-specific mortality. Reverse causation (disease causing pessimism) is unlikely because in that case both cardiovascular diseases and cancers would be expected to lead to pessimism.