This paper presents a rule-based Vietnamese-Thai machine translation (VTMT) system. Vietnamese text is input and segmented to a set of sentences through the use of punctuation marks. The output sentence is segmented into a sequence of words using a longest syllable matching algorithm together with named entity recognition (NER) rules. Segmented words will seek corresponding Thai words from the Vietnamese-Thai lexicon. Vietnamese to Thai transcription rules will be used to transcribe unknown words and recognized name entity words. The system will analyze the source sentence structure in order to generate the output Thai sentences. The translation accuracy obtained from the proposed system was 77.15%; which is better than the results achieved through the popular website Google Translate.