We compare two alternative definitions of the meridional surface temperature gradient (MTG) in the northern hemisphere. One focuses on equator-pole differences ((0¿--25 ¿N)-(65¿--90 ¿N)) and the other focuses on differences over midlatitudes ((30¿--35 ¿N)-(50¿--55 ¿N)). The MTG is sensitive to the choice of latitudes used to define it, and these two MTGs show substantial differences in behavior on timescales from interannual to multidecadal. The equator-pole MTG is plagued by missing data in the high-latitude belt, which affects its characteristics over at least the early part of the data record. Because the missing data are concentrated most heavily in the northernmost zones of the high-latitude belt, the variance of the equator-pole MTG is unduly high. This suggests a need to weight the equator-pole MTG by data availability in each latitude zone. The midlatitude-oriented MTG is more robust with respect to data availability and provides a measure of midlatitude circulation features. ¿ 1999 American Geophysical Union