There is a strong need for longitudinal and representative studies that focus on doctors' well-being. We conducted a longitudinal and nationwide cohort study (N = 631) of graduating medical students (T1), followed up in their first (T2), fourth (T3) and ninth (T4) postgraduate years. Response rates varied from 62 to 83%; the longitudinal response rate was 41%. We used comparison samples from a general population survey. Predictors of life satisfaction among the doctors were tested by linear regressions and repeated measures statistics. Doctors of both genders had significantly lower levels of life satisfaction than those in the comparison samples. Life satisfaction was lowest at T1, improved at T2, and stayed at the same level thereafter. Adjusted predictors of life satisfaction at T4 were lower age (P = 0.03), low level of neuroticism personality trait (P = 0.02), being married/cohabiting (P < 0.001), perceived social support (P < 0.001), lower work stress (P = 0.002), negative life events (P = 0.002), and frequency of physical training (P = 0.04). Only one-fifth of the total explained variance (R (2) = 0.34) was due to personality traits, whereas half was explained by married/cohabitant status and perceived social support. Being married/cohabiting, social support and physical training remained significant when mental distress (anxiety and depressive symptoms) was included in an additional multivariate model (R (2) = 0.49). Neuroticism trait (P < 0.001) was a significant predictor in the repeated measures model, but the increase in life satisfaction from T1 to T2 was predicted by lower levels of conscientiousness trait (P = 0.009). Norwegian doctors were less satisfied than an educationally age-matched group of the general population. Concurrent situational factors and stress, such as social support and mental distress were of major importance. However, personality clearly predicted life satisfaction over the years.