A detailed description of SQUID gradiometers is presented with the aim of improving their near field detection response while operating in unshielded, magnetically harsh urban environments. Spatial gradiometers of various orders are compared (0th, 1st, 2nd. and 3rd) and the effects on performance of constructional errors, near and far noise sources, and gradiometer baseline are examined. It is shown theoretically that the noise characteristics of a 3rd order spatial gradiometer are expected to be superior to that of lower order designs, particularly for noise sources located within intermediate distances (1.5–50 m) from the gradiometer. Following this analysis, a 3rd order SQUID gradiometer has been designed, constructed, and evaluated. Data are presented on the instrument noise characteristics for operation within an unshielded urban laboratory and examples of the 3rd order gradiometer response to various biomagnetic signals are shown. The results support the theoretical expectations and confirm that the 3rd order gradiometer exhibits better noise characteristics than a 2nd order gradiometer operating in similar environments.