Depressive symptoms are highly prevalent in psychotic populations and result in significant functional impairment. Limited knowledge of whether depressive symptoms are invariant across stages of illness curtails our ability to understand how these relate to illness progression. Clarifying the latent structure of depressive symptoms across stages of illness progression would aid etiological conceptualizations and preventive models. In the present study, one-factor (including all items) and two-factor (depression/hopelessness and guilt/self-depreciation) solutions were specified through confirmatory factor analysis (CFA). Measurement invariance analyses were undertaken across schizophrenia (SCZ; n = 312) and clinical high-risk (CHR; n = 175) groups to estimate whether the same construct is being measured across groups. Clinical correlates of the factors were examined. Results indicated that CHR individuals had a greater proportion of mood disorder diagnoses. Metric invariance held for the one-factor solution, and scalar invariance held for the two-factor solution. Notably, negative symptoms did not correlate with depressive symptoms in the SCZ group, though strong correlations were observed in CHR individuals. Positive symptoms were comparably associated with depressive symptoms in both groups. Results suggest depressive symptoms are more prevalent in CHR individuals. Targeting these symptoms may aid future efforts to identify risk of conversion. Further, some depressive symptoms may be systematically more endorsed in CHR individuals. Separating into depression/hopelessness and guilt/self-depreciation scores may aid comparability across stages of illness progression, though this issue deserves careful attention and future study.