The study of mRNA methylation is an emerging research field greatly fueled by recent advancement in high throughput sequencing technology. We propose here a binomial likelihood ratio test ('bltest') aiming at detecting differentially methylated mRNA with MeRIP-Seq data, 'bltest' models the read counts of each RNA methylation site in IP and input samples with a binomial distribution. If the successful rates of binomial distributions under two experimental conditions are significantly different, the corresponding RNA is identified to be differentially methylated. The proposed 'bltest' does not require an independent procedure to unify different library sizes of the MeRIP-Seq samples, thus retaining maximal information from the data. Our simulation results clearly showed that 'bltest' consistently outperforms classic 'Fisher's exact test' at all settings. The proposed method was also applied to the real data, and the findings are consistent with previous studies.