This paper presents unconstrained handwritten Kannada vowels recognition based upon invariantmoments. The proposed system extracts Invariant moments feature from zoned images. A Euclidiandistance criterion and K-NN classifier is used to classify the handwritten Kannada vowels. A total 1625image are considered for experimentation and overall accuracy found to be 85.53%. The novelty of theproposed method is independent of size, slant, orientation, and translation in handwritten characters.