A visually reinforced infant speech discrimination (VRISD) paradigm is described and evaluated. Infants at two ages were tested with the new paradigm on the following speech contrasts: [sa] vs [va], [sa] vs [fa], [sa] VS [za], [as] vs [a:s], vs [a:z], [at] vs [a:d], [a:t] vs [a:d], [at] vs [a:t], [fa] vs [thetaa]and [fi] vs [thetai]. The data reported are compared with data on the same speech contrasts obtained from three month olds in a high-amplitude sucking paradigm. Evidence suggesting developmental changes in speech-sound discriminatory ability is reported. Results are interpreted in light of salience of available acoustic cues and in terms of new methodological advances.