This study is the second phase of the author's larger attempt to investigate the factors affecting housewives' caregiving behavior to their elderly parents-in-law. Specifically, it revises and expands the previous model (1998) and develops a new one by rectifying the three major problems inherent in the previous study: (1) misspecification error; (2) non-equivalent comparison of results between the father-in-law model and mother-in-law model that stems from the inclusion of heterogeneous group of caregivers; (3) measurement problems for the two endogenous variables of eldercare attitude and behavior. To do this, the current study proposes a more comprehensive model by additionally incorporating other salient exogenous variables, renders the comparison of results between the father-in-law and mother-in-law models equivalent by including only homogeneous group of caregivers (i. e., only those housewives whose parents-in-law are both alive), and introduces standardized measurement scales for the endogenous variables. Estimation of the model in terms of maximum likelihood procedures in LISREL8 attests to a better overall performance over the previous model when judged from several criteria such as coefficient of determination, model fit statistics, proportion of significant causal paths, and measurement properties of reliability and validity for the variables. Interpretation of the findings suggests several salient theoretical implications that concern such crucial issues as the inconsistency between eldercare attitude and behavior, patterns of association among the subdimensions of eldercare, and the difference in the antecedents explaining attitude as opposed to behavior of eldercare. In particular, the finding that indicates almost no differences in the determinants between the father-in-law and mother-in-law models suggests a strong case to argue that caregiving behavior to fathers-in-law and mothers-in-law, respectively, is likely to be a uniform phenomenon sharing virtual1y the same antecedents, and that a unified single model is sufficient to account for caregiving behavior to both parties.