This paper examines the small-sample performance of spatial HAC (SHAC) estimators of the standard errors on parameters. We find that, in small to moderately-sized datasets, the use of HAC estimators may be recommended only with a relatively large degree of cross-sectional interdependence.