In order to determine what ground-based proxies are best for evaluating solar irradiance variation before the advent of space observations, it is necessary to test these proxies against space observations. We have tested sunspot number, total sunspot area, and sunspot umbral area against the Nimbus-7 measurements of total solar irradiance variation over the eleven year period 1980-1990. The umbral area yields the best correlation and the total sunspot area yields the poorest. Reasons for expecting the umbral area to yield the best correlation are given, the statistical procedure followed to obtain the results is described, and the value of determining the best proxy is discussed. The latter is based upon the availability of an excellent database from the Greenwich Observatory obtained over the period 1876-1976, which can be used to estimate the total solar irradiance variation before sensitive space observations were available. The ground-based observations used were obtained at the Coimbra Solar Observatory. The analysis was done at Goddard using these data and data from the Nimbus-7 satellite.