Diagnosis of Pythium keratitis is problematic due to the difficulty in obtaining a culture report resulting in unnecessarily prolonged usage of antimicrobial medication due to misdiagnosis. This study evaluated and compared nested PCR technique with culture and immunoperoxidase staining assays of Pythium insidiosum in paraffin-embedded corneal tissues from patients with suspected fungal keratitis. Six of 51 pathological reports compatible with fungal infection and 6 of 48 culture-proven fungal keratitis were identified as Pythium. Twenty-seven specimens were PCR-positive for Pythium insidiosum. In comparison with fungal culture for P. insidiosum, PCR had 83% sensitivity and 77% specificity with fair agreement (Kappa score of 0.227, p = 0.001). The mean age of PCR-positive is younger than PCR-negative group and there is a female preponderance in Pythium-infected group (p = 0.002 and p = 0.004, respectively). Nineteen specimens had positive results using immunoperoxidase staining assay with fair agreement to culture method (Kappa 0.340, p < 0.001), and 83% sensitivity, 85% specificity and 85% accuracy (95% CI: 76.7-90.7). PCR-based technique compared with culture and/or immunoperoxidase staining assay had 91.7% sensitivity, 81.8% specificity and 83% accuracy (95% CI: 74.5-89.1) with moderate agreement (Kappa 0.477, p < 0.001). Thus nested PCR detection of P. insidiosum should be employed in preliminary diagnosis of Pythium keratitis in order to initiate proper management.