If solar power is to provide substantial portions of our electricity needs, it will first become cost effective when it provides peak power in the daytime, without the need for storing the energy. Indeed since human electricity consumption is frequently small at night and larger when the sun is shining, there is already a natural correlation. Existing power systems are currently geared to provide this variable demand, with baseload plants cheaply providing a constant level of power, and dispatchable plants dynamically (and more expensively) supplying the rest. This leads to the frequent suggestion that one can exploit the correlation between sunlight and electricity by using energy from solar panels during the day to offset some of the load previously generated by dispatchable plants. This paper addresses the question of how much of the load can be substituted by the solar electricity, without leaving the solar power plant substantially idle or requiring the solar power to be stored. It uses historical sunlight and electrical load data from 32 regions of the United States to determine the photovoltaic (PV) power generation capacity that could be installed such that "almost all" of its energy output would occur at times of high demand. Specifically, what is the maximum deployment that permits 95% of the annual output from PV to be utilized without reducing the output of the baseload plants? Our results for these 32 regions are that 7.8% of the total annual electricity demand could be met by installing 59 GW of PV panels. This represents about a fourth of the present electrical energy supplied by dispatchable plants. If solar power were equally effective in the rest of the United States, nearly 200 GW of PV capacity could be put to use without any energy storage. Thus, in the near term, there is enormous room for expanding the roughly 1 GW installed base of PV power without investing in night-time energy storage. The paper also provides insight into how year-to-year variability of sunlight and demand impact the results. (C) 2009 Elsevier Ltd. All rights reserved.