We have examined the response of phage T4 nonsense mutations located at various sites within the same cistron to different suppression agents. A wide range of suppression efficiency is found for both ochre (UAA) and amber (UAG) mutations under conditions where suppression provides a measurement of the amount of chain propagation past the mutated site. We have established a relationship between our measurement-the size of the phage yield-and the amount of rIIB product present in the infection. Our data suggest that the 1000-fold range of variations in yields observed in the rIIB cistron corresponds to a 30-fold range of variation in the level of rIIB product, i.e. in the relative frequency of chain propagation past the various nonsense codons included in our test. From the parallelism of response of any particular mutant to very different suppression mechanisms we conclude that the efficiency of suppression is site specific, that is to say, that the main factor determining the frequency of chain propagation at a nonsense codon by any type of suppression mechanism is the nucleotide sequence adjacent to the nonsense codon (reading context). We propose that the recognition of a natural termination signal involves a sequence longer than a nonsense codon and that nonsense codons outside of their natural environment induce variable termination rates which are reflected in the suppression potential.