Reliable reference genes are critical for relative quantification using quantitative real-time PCR (qPCR). Ten tomato genes (Solanum lycopersicum) and their respective primer sets, which have been used over the last 6 years as references in expression studies, were evaluated for their performance using leaf tissue samples grown under semi-controlled conditions and infected with grey mould (Botrytis cinerea) or late blight (Phytophthora infestans). The target genes coding for U6 snRNA-associated Sm-like protein LSm7, calcineurin B-like protein and V-type proton ATPase were the most stable expressed of all the genes tested in three experimental repetitions. Evaluation of candidate reference genes with geNorm and NormFinder softwares yielded the lowest mean values for their respective primer sets LSM7, SlCBL1 and SlATPase, suggesting stable expression. However, SlATPase primer set revealed a comparably high intra-group variation and was thus not considered further. In follow-up experiments with P. infestans, the geNorm and NormFinder values of primer sets LSM7 and SlCBL1 were even lower, indicating the stability of their expression also under these conditions. Primer efficiency differed by -18 to +5 percentage points from values presented in the literature. Our findings show that a reference primer set which delivers the best results in one system may be outperformed by another under different experimental conditions, thus recommending a reassessment of both expression stability and qPCR efficiency whenever the biological or technical experimental set-up is changed. On the basis of our results, we recommend the use of LSM7 and SlCBL1 as reference primer sets for gene expression studies on plant tissue derived from open or semi-controlled conditions.