A quasi-synoptic study of the shelf water/slope water front off New Jersey, Delaware, and Maryland in mid-July 1977 revealed an isolated body of very cold (<6 ¿C) water in the near-bottom 'cold pool'. Its volume at least halved over 10 days during which nearby shelf waters were perturbed by two anticyclonic (warm core) eddies located over the continental slope. Water colder than 6 ¿C lay along the bottom between the 80- and 100-m isobaths and was about 30 m thick. The seaward edge of the cold pool (here denoted as colder than 8 ¿C) marked the inshore boundary of the shelf water/slope water subsurface temperature front, with 10 ¿C variations per 2--20 km, depending upon alongshelf location. The thermocline underwent large displacements and deformations over 10 days, indicative of vigorous upwelling and downwelling and large potential vorticity changes at the shelf break. The dominant scales of variability (or correlation scales) in the thermal fields were 30 km alongshelf, 10 km cross shelf, and 10 to perhaps 20 days. (The internal radius of deformation on the outer shelf is about 15--25 km.) There was substantial vertical structure on the scale of 10 m or less. A surface convergence zone coincided with a surface rougness band ('tide rip') near the shelf break, and internal gravity wave activity seemed intense there too. Two independent estimates of vertical velocity at the shelf break suggested values of about 10-2 cm/s, one for a time scale of a week or longer and the other for a time scale of at least a few hours. Over Wilmington Canyon a large 'lip' of cold pool intruded into the slope water along the southwestern side of the Canyon, perhaps undergoing entrainment and 'calving' by an anticyclonic eddy located over the slope. There was further evidence of offshore entrainment throughout the water column between the two eddies.