Background: Information on cancer stage at diagnosis is critical for population studies on cancer care and outcomes. Most population-based cancer registers include a substantial proportion of unstaged cancers, and few studies have examined factors impacting on staging or outcomes for unstaged patients. This study investigated prevalence of unstaged cancer in New Zealand, explored factors which predict unstaged disease, and described the outcomes of unstaged patients. Methods: Patients diagnosed with the most prevalent 18 cancers between 2006 –2008 (N = 41,489) were identified from the NZ Cancer Registry, with additional data obtained from mortality and hospitalisation databases. Logistic and Cox regression were used to investigate predictors and outcomes of unstaged disease. Results: Three distinct groupings of cancers were found based on proportion of unstaged patients (low = 2-14% unstaged; moderate = 33-42%; high = 65-73%). Increasing age was a significant predictor of an unstaged diagnosis across all groups (adjusted Odds Ratios [OR]: 1.04-1.05 per year of age across groups). Māori patients were significantly more likely to be unstaged for cancers with a low (OR=1.21 [95 % Conf. =1.01 - 1.44]) or moderate proportion (OR=1.32 [1.17 - 1.50]) of unstaged patients. Patients with comorbidity were significantly more likely to be unstaged only in those cancers with a low proportion of unstaged patients (OR=2.25 [1.85 - 2.74]). Likelihood of receiving definitive surgical treatment differed by cancer, with no clear pattern by proportion of unstaged patients. Unstaged patients tended to have better one-year survival than those with distant disease, but poorer survival than those with regional disease. Although this order was consistent, the magnitude of the difference in mortality hazard between these stages differed considerably across groups. Conclusion: Both the characteristics of unstaged cancer patients and their outcomes vary depending on the prevalence of unstaged cancer within a given cancer site.