The objective of this study was, firstly, to determine the factor structure and factor invariance of the Zung Self-Rating Depression Scale (ZSDS) and, secondly, to justify its use in coronary patients (CPs) and healthy people (HP). Two comparable samples of males were studied: 217 CPs and 191 HP. Exploratory and confirmatory factor analyses (EFA and CFA) for ordinal data were carried out with Mplus. Two models obtained from all participants in this study and another two, the model of Shafer (2006) from a meta-analysis and that of Barefoot et al. (2000) with CPs, were analyzed in CFA. A two-factor structure was supported by EFA in both samples, but none of the models showed adequate goodness-of-fit for the CPs and the HP in CFA. Only the two and three-factor models obtained from the combined sample of CPs and HP showed adequate goodness-of-fit for HP. The ZSDS showed good reliability, replicated the prevalence of depressive symptoms found in other studies and was able to distinguish between CPs and HP. We conclude that the best fit is obtained from the two-factor solution in HP, that the factor structure of the ZSDS is not invariant and is linked to positively and negatively worded items.