The aim of this study was to advance the imputation of missing values for some autoregressive moving average models (ARMA) with generalized autoregressive conditioned heteroscedastic (GARCH) models.