This is a qualitative descriptive research, which is to find common errors on the students? narrative writing on the 2nd grade students of SMA Negeri 95 Jakarta. The research was conducted on July 24th, 2008. The instrument of the research is the writing task. The writer uses descriptive method which describes the analysis data based on the discovery. Population of the respondents was the 2nd grade students of SMA Negeri 95 Jakarta. The sample of the respondents was the students of XI IPS 4. This study was a non-experimental study. The study referred to the approach of error analysis. The procedures of it included identifying the corpus, classifying the corpus, evaluating the error. The research activities were giving the writing task to the students, asking the respondents to write a narrative writing based on the writing task given, analyzing the errors which the students made, and revising the writing errors. The scope and limitation of the research was on the grammatical aspects based on Betty Schramfer Azar?s error guideline of students? writing and sources of errors stated by Richards. The grammatical aspects which were analyzed are: singular-plural, word form, word choice, verb tense, add a word, omit a word, word order, incomplete sentence, spelling, punctuation, capitalization, article, meaning not clear, run-on sentence. The sources of error which were analyzed are interlanguage error, overgeneralization, ignorance of rule restriction, incomplete application of rules, false concept hypothesized. The result of the research revealed that common errors on the students? writing ability were the error of verb tense takes the first place with 19.88 % followed by the errors word choice with 15.33 %, the errors of add a word with 13.05 %, the errors of punctuation with 12.60 %, the errors of capitalization with 9.71 %, the errors of article with 9.26 %, the errors of omit a word with 9.11 %, the errors of spelling with 4.70 %, the errors of word order with 1.67 %, the errors of incomplete sentence with 1.52 %, the errors of singular-plural and meaning not clear with 1.21 %, the errors of word form with 0.75 %. Then from the calculations of the sources of errors, it can be seen that ignorance of rule restriction takes the first place with 59.18 % followed by overgeneralization with 21.54 %, interlanguage error with 15.32 %, incomplete application of rules and false concept hypothesized with 1.98%.