The paper discusses the application of differential pulse-code modulation (d.p.c.m.) and hybrid d.p.c.m./p.c.m. coding as a means of reducing the bit rate required for broadcast-quality System I (Pal, 625-line, 5.5 MHz) video signals. In the d.p.c.m. codes investigated, each codeword indicated the difference in magnitude between a sample of a video signal and a prediction of this sample based on samples already transmitted in the same and/or previous line periods with a sampling frequency equal to three times the colour subcarrier frequency. The codes examined required the transmission of between 3 and 6 bits per sample. Subjective test results are given for the picture impairment caused by quantising errors and transmission errors. The tests on quantising errors indicated that d.p.c.m. could provide a saving of nearly 2 bits per sample compared with p.c.m. A comparison of subjective test results and measurements of the r.m.s. quantising errors for d.p.c.m. and p.c.m. indicates that the picture impairment corresponding to a given signal/noise ratio is similar for both d.p.c.m. and p.c.m.