Trivariate vector error-correction models (VECMS) in conjunction with cointegrated systems, are used to examine the causality issues between U.S. employment and energy consumption. Results of the VECMs indicate a bidirectional causality between nonfarm employment and energy consumption and between total employment and energy consumption. Furthermore, the replication of prior studies, after log-transforming, adjusting for seasonality and correcting for unit roots, also produced a unidirectional causal relationship from total employment to energy consumption and a bidirectional causal relationship between nonfarm employment and energy consumption for the 1973.1-1984.6 sample period in the bivariate models. Extending the sample period further, however, completely eliminated the unidirectional causal relationship from total employment to energy consumption in the bivariate system.