Thermography is widely used for inspecting electrical systems where costly problems are often preceded by telltale thermal signatures. Many thermographers, or their customers, however, mistakenly rely on radiometric temperature data to prioritize these findings. Due to the inherent limitations of radiometry and the complexities of heat transfer in the components being inspected, the data is all too often either inaccurate or misunderstood. Routinely, however, thermographers proceed as if nothing is amiss. This is due to an understandable, but misguided, attempt to simplify the decision-making process regarding repair priorities. The result is that predictions of repair priorities are not as accurate as they could be. On the one hand failures still occur while on the other repairs are often made inappropriately. In this paper we discuss the problems encountered when collecting and interpreting radiometric data. We will also outline a simple, effective system that thermographers are using to dramatically improve the predictive value of the technology. Improved results are achieved first separating the often intertwined questions of 'when will the component fail?' and 'what will be the consequences of failure?' The second improvement comes from incorporating all relevant data in the decision making process and weighing the impact each has. The system described, which can easily be adapted to diverse needs, is being used successfully and with repeatable results that have been shown to improve with usage.