Templating of preoperative X-rays is routinely performed before a total knee replacement. We performed this retrospective study on preoperative templating in primary total knee replacement to assess its accuracy and reproducibility. Preoperative radiographs of 47 patients were templated twice by two observers separately after a gap of 3 months. The size of actual implant used was compared to the size predicted by templating. Data were analysed to assess inter and intra-observer variation. Templating was accurate for both tibial and femoral components in only 53.2% of observations. This study also revealed that templating is a highly-subjective and observer-dependent technique. Inter-observer and intra-observer mismatch was present in 46.8% and 43.6% of readings respectively. We conclude from this study that preoperative templating is neither accurate nor reproducible.