This paper investigates the practical consequences of using fully conditional specification (FCS) in multiple imputation for incomplete multivariate data. FCS involves specifying separate conditional densities for each variable given the other variables, which can lead to incompatible models if the specified densities do not form a consistent joint distribution. The study uses simulation to evaluate the performance of FCS under four different missing data mechanisms: missing completely at random (MCAR), missing at random (MAR), missing not at random (MNAR), and mixed mechanisms. The results show that multiple imputation using FCS produces unbiased estimates with appropriate coverage even under incompatible models. The method is computationally efficient, requiring only a few Gibbs sampling iterations, and performs well in both univariate and multivariate settings. The study also examines the robustness of FCS under incompatible models, finding that it remains effective in producing reasonable imputations. Overall, the paper concludes that FCS is a robust and flexible approach for handling missing data, despite its theoretical limitations.This paper investigates the practical consequences of using fully conditional specification (FCS) in multiple imputation for incomplete multivariate data. FCS involves specifying separate conditional densities for each variable given the other variables, which can lead to incompatible models if the specified densities do not form a consistent joint distribution. The study uses simulation to evaluate the performance of FCS under four different missing data mechanisms: missing completely at random (MCAR), missing at random (MAR), missing not at random (MNAR), and mixed mechanisms. The results show that multiple imputation using FCS produces unbiased estimates with appropriate coverage even under incompatible models. The method is computationally efficient, requiring only a few Gibbs sampling iterations, and performs well in both univariate and multivariate settings. The study also examines the robustness of FCS under incompatible models, finding that it remains effective in producing reasonable imputations. Overall, the paper concludes that FCS is a robust and flexible approach for handling missing data, despite its theoretical limitations.