Preprint of paper published in: ENASE 2008 - Proceedings of the 3rd International Conference on Evaluation of Novel Approaches to Software Engineering, 4-7 May 2008 Fault screeners are simple software (or hardware) constructs that detect variable value errors based on unary invariant checking. In this paper we evaluate and compare the performance of two low-cost screeners (Bloom filter, and range screener) that can be automatically integrated within a program, while being automatically trained during the testing phase. While the Bloom filter has the capacity of retaining virtually all variable values associated with proper program execution, this property comes with a much higher false positive rate per unit training effort, compared to the more simple range screener, that compresses all value information in terms of a single lower and upper bound. We present a novel analytic model that predicts the false positive and false negative rate for both type of screeners. We show that the model agrees with our empirical findings. Furthermore, we describe the application of both screeners, where the screener output is used as input to a fault localization process that provides automatic feedback on the location of residual program defects during deployment in the field.