RT-PCR is more sensitive for rhinovirus detection than cell culture, but healthy controls are frequently rhinovirus (or picornavirus) positive in cross-sectional studies. Fifteen healthy children were followed over at least three seasons of the year with weekly sampling of nasal/nasopharyngeal secretion for RT-PCR testing for picornavirus and daily recording of respiratory symptoms. One sample positive for picornavirus was diagnosed as an infection; consecutive positive weekly samples constituted a single infection. Picornavirus illness was diagnosed if RNA was detected 7 days prior through 21 days after onset. One hundred fifty-five (21%) of 740 weekly samples were picornavirus positive and associated with illness; 37(5%) positives were not associated with illness (P = 0.001). The 192 positive samples occurred in 121 infections, 74 with a single positive and 47 with "runs" of positives in two or more consecutive samples. Forty five (96%) of the 47 runs comprised 2 or 3 consecutive positives. Ninety six (52%) of 185 reported illnesses during 235 child-months were picornavirus positive (0.4/child-month); 25 infections were asymptomatic (0.11/child-month). The infection rate was highest in fall (0.66/child-month); the winter rate (0.44/child-month) was similar to that in spring (0.5) and summer (0.43). Picornavirus infections in healthy children were common (0.51/child-month), episodic, and usually associated with brief illness; one fifth of infections were asymptomatic. The infection rate was highest in fall; infections in winter occurred at the same rate as in spring and summer.