Te aim of the present research was to identify environmental features and personal characteristics and their relationship to mental health among long-term unemployed people in New Zealand. Warr's Vitamin Model of unemployment was investigated in a cross-sectional/longitudinal study. In the Main Study 532 people who had been unemployed for at least six months, were surveyed. A wide ranging questionnaire, the General Health Questionnaire 12 (GHQ) and the Rosenberg (1965) Self-Esteem Scale (RSE) were administered. This was followed by an interview in which people selected themselves into 4 groups. One group or 28.6% "wanted employment", 35% felt "not able to work", 27.2% stated having "alternatives to employment" and a small minority of 9.2% were "interested in training". One year later, people from each of the 4 unemployed groups and one group of re-employed people were re-interviewed in the Follow-up Study (N=99). A shortened questionnaire, the GHQ, the RSE and the Hopkins Symptom Checklist 25 (HSCL) were administered. A series of multiple regressions supported Warr's Vitamin Model, as five environmental features were associated with mental health levels in the Main Study and two in the Follow-up Study. A number of personal characteristics were also significant predictors of mental health. Generally, mental health levels of the unemployed were low, but re-employment resulted in significant improvements. The results of the Follow-up Study showed that the percentage of people "wanting employment" and "interested in training" decreased, whereas the group having "alternatives to employment" increased. The group "not able to work" stayed the same in size. Mental health levels were low in the groups "wanting employment" and "interested in training". Extremely low mental health was evident in the group "not able to work". In the group "alternatives to employment" mental health levels were high and more similar to levels observed in the employed group. Several moderator variables of the negative experience of unemployment were tested such as "age", "gender", "marital status" and "stated ethnic origin". The results show that long-term unemployed people are not a homogeneous group and significant differences in mental health exist within sub-groups. Recommendations for future research were made.