This study investigates the relationship between agricultural electricity consumption and agricultural output for a panel of 12 regions of Turkey for the period 1995-2013. In order to reveal the possible heterogeneity between regions, empirical analyses are conducted for the whole panel data and two subgroups within the panel data; namely, coastal regions and non-coastal regions. The results from several panel unit root tests indicate that electricity consumption and output are stationary process at their levels for overall panel and the two specific groups. By using the OLS with regional fixed effects, this study finds that coefficient estimate of electricity consumption on output is statistically significant and positive for overall regions, coastal regions and non-coastal regions. In addition, the results from the Dumitrescu-Hurlin Granger causality test show that there is unidirectional causality running from agricultural output to electricity consumption for non-coastal regions, and there is bidirectional causality between agricultural electricity consumption and output for overall panel and coastal regions. Findings and policy implications are further discussed. (C) 2016 Elsevier Ltd. All rights reserved.