This thesis aims to examine the relationship between the psychosocial working conditions of hospital doctors in England and the quality of care being provided, with work-related wellbeing as a mediator. It applied the job demands-resources model to this occupational sample, and utilised a multilevel perspective to include trust-level demands and outcomes. In the first of four studies - a systematic review and meta-analysis found that across 21 studies, 62% of the reported relationships between job demands and 64% of job resources’ relationships with quality of care were significant; the presence of these relationships varied by the type of outcome measure used. A lack of theoretical grounding within these studies emphasised the need to frame these relationships within a theoretical framework. The three subsequent empirical studies drew on composite scales from the 2014 NHS Staff Survey in England. Across these multilevel studies, job demands (insufficient work resources, workplace aggression) predicted negative work-related wellbeing (presenteeism, work-related stress), while job resources (manager support, job control, effective teams) predicted work engagement. Trust-level demands (number of emergency admission, bed occupancy rate) also predicted hospital doctors’ work-related wellbeing. No interactions were observed between job demands and resources. Work-related wellbeing mediated most relationships between job demands and resources with individual self-rated quality-of-care measures. Some mediations involving patient satisfaction with doctors were found, but not for hospital mortality or patient safety incidents. The research reported in this thesis highlights the complexity of work-related predictors to hospital doctors’ work-related wellbeing and the quality of care provided. It further demonstrates that these outcomes are a product of their wider work context. Successful interventions should target the appropriate antecedent pathway, and recognise trust and system factors. The job demands-resources model can be useful in explaining individual-level relationships, but is limited when including trust-level measures. Further implications on research, practice, and policy are discussed.