The core element of many existing approaches to face detection is the classification algorithm that determines if a sub-image of an input image contains a face pattern. In this paper, we present a novel and effective distribution-based face/non-face classification technique that detects frontal face patterns with possible in-plane rotation. A 15x15 input sub-image is first processed by a color filter, which verifies the presence of human skin color in the sub-image. Then, the intensity image is extracted from the identified skin color sub-image and converted into a vector in a high-dimensional space (R225). Principal component analysis is employed to reduce the dimension of this space to 20. In our approach, the distributions of face and non-face patterns in the R20 space are modeled using mixtures of Gaussians. The parameters of the Guassian mixture models are determined through the use of the Expectation/Maximization (EM) algorithm. Finally, the classification of sub-images into face or non-face patterns is carried out through comparison of their estimated probability density functions. Experimental results have shown that the proposed technique is capable of performing highly accurate face/non-face classification.