We propose a temporally adaptive 3-D subband coding (3-D SBC) technique to effectively exploit the temporal activities in the input image sequence. By using the rate-distortion (R-D) performance measure, we show the optimal number of temporal subbands can be easily determined. The base temporal subband, which yields much concentrated energy, is encoded using the H.261-like motion-compensated discrete cosine transform (MC-DCT) technique. While in the higher temporal subbands, the two-dimensional (2-D) adaptive wavelet packet bases are employed to exploit the various energy distributions, due to the moving components. In encoding the subbands, we employ adaptive scanning methods, followed by uniform step-size quantization with variablelength coding (VLC), and a coded/not-coded flag reduction technique based on the quadtree structure. From the simulation results, the proposed 3-D SBC provides about 0.29- to 3.14-dB PSNR gain over the H.261 technique and the temporally fixed 3-D SBC techniques at bit rates 300 to 512 Kbits/s.