Purpose: To examine the prevalence of $Clostridium$ $difficile$ ($C.$ $difficile$) colonization (CDC) and potential neonatal determinants of CDC in hospitalized preterm infants. Methods: Fecal samples were serially collected within 72 h after birth and at 1, 2, and 4-6 weeks of age from preterm infants in the neonatal intensive care units (NICUs) of two different university hospitals. Total bacterial DNA was extracted from each fecal sample from 49 infants, and polymerase chain reaction (PCR) was performed with primers for the 16S gene of $C.$ $difficile$ and the toxin A and toxin B genes. The correlation between the results of $C.$ $difficile$ PCR assays and the clinical characteristics of the infants was analyzed. Results: The prevalence rates of CDC were 34.7, 37.2, 41.3, and 53.1% within 72 h after birth and at 1, 2, and 4.6 weeks of age, respectively. The toxin positivity rate was significantly higher in the infants with persistent CDC than in those with transient CDC (8/12 [66.7%] vs. 6/25 [24.5%] ($p$=0.001). Among the various neonatal factors, only the feeding method during the first week after birth was significantly associated with persistent CDC. Exclusive breast-milk feeding (EBMF) significantly decreased the risk of persistent CDC compared to formula or mixed feeding (adjusted odds ratio: 0.133, 95% confidence interval: 0.02-0.898, $p$=0.038). Conclusion: The prevalence of CDC increased with the duration of hospitalization in preterm infants in the NICU. EBMF during the first week after birth in hospitalized preterm infants may protect against persistent CDC.