We conduct a case study on applying an approach called reverse pre-reordering (REV-REO) to Japanese-to-Khmer (JKm) statistical machine translation (SMT). The REV-REO is a simple but efficient approach originally designed for Japanese-to-English SMT to resolve the word ordering problem in translation. Theoretically, REV-REO can be applied on translation tasks from a head-final language to a head-initial language. As Khmer is a more typical head-initial language than English is, REV-REO should have effects on JKm SMT. We conducted experiments of JKm SMT using a standard phrase-based SMT system with a Japanese-Khmer parallel corpus of more than 100 thousand sentence pairs. The performance of REV-REO is evaluated and investigated by three automatic measures. The experimental results suggest that established natural language processing techniques can work well on the understudied Khmer language.