Twenty healthy preterm infants (gestational age 26 to 33 weeks, postmenstrual age [PMA] 32.1 to 39.6 weeks, postnatal age [PNA] 2.0 to 11.6 weeks) were studied weekly from initiation of bottle feeding until discharge, with simultaneous digital recordings of pharyngeal and nipple (teat) pressure and nasal thermistor and thoracic strain gauge readings. The percentage of sucks aggregated into 'runs' (defined as > or = 3 sucks with < or = 2 seconds between suck peaks) increased over time and correlated significantly with PMA (r=0.601, p<0.001). The length of the sucking-runs also correlated significantly with PMA (r=0.613, p<0.001). The stability of sucking rhythm, defined as a function of the mean/SD of the suck interval, was also directly correlated with increasing PMA (r=0.503, p=0.002), as was increasing suck rate (r=0.379, p<0.03). None of these measures was correlated with PNA. Similarly, increasing PMA, but not PNA, correlated with a higher percentage of swallows in runs (r=0.364, p<0.03). Stability of swallow rhythm did not change significantly from 32 to 40 weeks' PMA. In low-risk preterm infants, increasing PMA is correlated with a faster and more stable sucking rhythm and with increasing organization into longer suck and swallow runs. Stable swallow rhythm appears to be established earlier than suck rhythm. The fact that PMA is a better predictor than PNA of these patterns lends support to the concept that these patterns are innate rather than learned behaviors. Quantitative assessment of the stability of suck and swallow rhythms in preterm infants may allow prediction of subsequent feeding dysfunction as well as more general underlying neurological impairment. Knowledge of the normal ontogeny of the rhythms of suck and swallow may also enable us to differentiate immature (but normal) feeding patterns in preterm infants from dysmature (abnormal) patterns, allowing more appropriate intervention measures.