Purpose: Karyotype analysis in hereditary retinoblastoma is considered to be of marginal value in risk prediction due to uncertainities in the assessment of 13q14 deletions. However, it is a low cost genetic test for retinoblastoma in developing countries. In the present study, the results of karyotype analysis were refined by a statistical method to overcome limitations. Methods: Karyotype analysis was performed by trypsin - Giemsa banding and naked eye karyotyping for 33 bilateral, 25 unilateral and one regressed retinoblastoma patients. The percentage of metaphases with 13q14 deletions in each case was plotted on a scatter diagram. Normalization of the data was achieved by log transformation and the results were statistically analyzed by one-sample‘t’ test using SPSS version 9.0. Results: Seven samples had 13q14 deletion percentages above the cutoff value. One-sample‘t’ test showed significance (p< 0.001). By this method, two unilateral and five bilateral patients had 13q14 deletions, constituting 11.8 % of cases. Conclusion: For accuracy, statistical analysis should be considered as an adjunct in karyotyping.