The goal of this investigation was to characterize the manual and automated response of residential customers to high-price "critical" events dispatched under critical peak pricing tariffs tested in the 2003-2004 California Statewide Pricing Pilot. The 15-month experimental tariff gave customers a discounted two-price time-of-use rate on 430 days in exchange for 27 critical days, during which the peak period price (2 p.m. to 7 p.m.) was increased to about three times the normal time-of-use peak price. We calculated response by five-degree temperature bins as the difference between peak usage on normal and critical weekdays. Results indicated that manual response to critical periods reached -0.23 kW per home (-13%) in hot weather (95-104.9°F), -0.03 kW per home (-4%) in mild weather (60-94.9°F), and -0.07 kW per home (-9%) during cold weather (50-59.9°F). Separately, we analyzed response enhanced by programmable communicating thermostats in high-use homes with air-conditioning. Between 90°F and 94.9°F, the response of this group reached -0.56 kW per home (-25%) for five-hour critical periods and -0.89 kW/home (-41%) for two-hour critical periods.