Formally called ITU-T Recommendation H.261, this motion video compression algorithm was developed in 1990 to provide video conferencing over ISDN telecommunications networks. This algorithm was designed to operate at a constant data rate at rates from 56 Kbps to 2 Mbps. It supports two resolutions: Common Interchange Format (CIF) which is 352×288 pixels and Quarter Common Interchange Format (QCIF) which is 176×144 pixels. Both resolutions use the 4:2:0 YCbCr chroma sampling scheme.

H.261 was the first practical digital motion video compression standard. It divided the image into macroblocks consisting of a 16×16 array of luma samples and an 8×8 array of chroma samples. Inter-picture prediction is used to reduce the temporal redundancy within the video. Motion vectors are used to account for macroblock motion between the actual and predicted image. Discrete Cosine transform (DCT) is then used on the five 8×8 blocks within each macroblock of the predicted image. Scalar quantization is used on the transform coefficients and Variable Length Coding (also called Entropy Coding) is used to remove statistical redundancy in the resulting values. Some number of Intra-coded (non predicted) macroblocks are distributed through each frame, such that over a given period of time, every macroblock in the image has been intra-coded. This is done to refresh transmission and mathematical round-off errors.

 These techniques form the basis of many or the motion video compression algorithms that have been developed since H.261.