We have developed an image database of handwritten (off-line) basic characters of Devenagari, the most popular script of India. The present database consists of approximately 30000 samples of handwritten isolated basic characters. However, these samples are not evenly distributed over 49 possible classes. The main reason of such uneven distribution is that these data were prepared automatically by using a software from certain form type documents. The subjects had been asked to write individual isolated character of Devanagari on separate rectangular box preprinted on this form type document. This software failed in several situations and the failure cases were excluded from the database through manual checks. A few samples from this database are shown below.