Due to the rapid increase in deployments and high penetration of solar power generation worldwide, solar power generation forecasting became critical to integration planning, operations and trading of solar power. Utilities and electricity system operators require day-ahead and hour-ahead as well as intra-hour solar power forecasts for operational reasons, solar power producers and energy traders are also interested in high quality solar power forecasts. The proposed paper is dedicated to technology roadmap of solar power forecasting and the results of field performance evaluation of the SolarSatDataTM set of technologies at utility-scale PV power plants in California in 2010-2011. The paper discusses the major forecasting time horizons required by the power industry including intra-hour forecasting (5-min rolling 15 minutes ahead forecasts), hour ahead and day ahead forecasts as well as longer-term (7+ day and month ahead) forecasting requirements. Applicability of leading technologies to required forecast horizons and their efficiency in terms of solar power forecast precision is discussed. Reviewed technologies include persistence forecasting, numeric weather predictions, satellite imagery processing and sky imagers. The paper presents approaches to solar forecasting performance evaluation. Because of the erroneously perceived simplicity of solar radiation forecasting, very often non-repeatable, poorly explained or obscure estimates of solar power forecast performance are used. This creates uncertainty with the quality of forecasting service as well as unrealistic expectations of possible forecast precision. Because of these circumstances there is a need for defining a common methodology for evaluating forecast performance, establishing verification procedures and setting common standards for industry-approved quality of solar forecast performance. Solar power forecast quality claims can be easily verified when the source of forecast is known. Most often the offered power generation forecasts are based on publically available results of Numerical Weather Prediction (NWP) models and on the use of empirical relationships between solar resource and generated power at a specific plant. The quality of these forecasts is limited by the quality of the used NWP models which is known. Less frequently the solar radiation is estimated based on proprietary models such as satellite - based or total sky imager – based cloud cover and radiation forecasts. In such cases there are also known limits to the accuracy of prediction which help objectively evaluate forecast performance precision claims. This paper is proposing a set of standards for evaluating intra hour, hour ahead, day ahead and week ahead solar power forecast performance. The proposed standards are based on sound methodologies and extensive field practice and offer a solid ground for reliable inter-agency comparisons of forecast performance. The discussed approach to evaluating field performance of various forecasting technologies is supported by the results of day-ahead and hour-ahead solar power forecasts in three utility service areas in California.