1. The presence of a direct lateral geniculate nucleus (LGN) input to area V2 raises the possibility that some neurons in this area remain active when area 17 is inactivated. It is also known that many neurons in area MT are visually responsive in the absence of input from area 17. Because MT sends a strong projection to V2, it appears likely that visual activity could be transferred to V2 through this feedback connection when the V1 afferents are disabled. For these reasons, we decided to reexamine the residual visual activity of neurons in V2 during inactivation of area 17. A circular region 16 mm in diameter on the opercular part of area 17 was cooled by a thermoelectric Peltier device, and single- and multiunit activity was recorded in the retinotopically corresponding region of area V2. 2. Because of the proximity of areas V1 and V2, it was necessary to make sure that neurons in V2 could not be directly blocked by cooling applied to V1. Temperature gradients within cortex were measured with a specially designed thermocouple at different heat flows imposed by the Peltier device. Gradients ranged between 2.3 and 5.5 degrees C/mm. Knowing the temperature gradients and the temperature of the cooling plate, it was possible to deduce the temperature at a given depth within cortex. With this method we measured the blocking temperatures of neurons in area 17, i.e., the temperature at which neurons completely ceased to respond to optimal visual stimulation. Blocking temperatures ranged between 4 and 18 degrees C, values that are substantially lower than those reported in previous papers. Knowing the blocking temperatures, it was possible to determine the cooling-plate temperature necessary to entirely block the region of V1 under the cooling plate. Using the temperature gradients, we then calculated the depth of recording for which V2 neurons could not be directly blocked by the cold. For this reason, all our recordings were made in or near the fundus of the lunate sulcus. 3. During cryoblocking of V1, we recorded 154 sites in penetrations normal to area V2. All these sites had receptive fields included within the visual-field region coded in the cooled zone. In addition, we recorded 55 sites in tangential penetrations aimed at traveling in V2 for long distances. Among these 209 sites, only 3 could be considered as unambiguously active when V1 was blocked. Two of these sites were located at or close to the V2-V3 border.(ABSTRACT TRUNCATED AT 400 WORDS)