The relationship of mutations between 241 H5N1 hemagglutinins of influenza A virus was determined according to their amino acid and RNA codon sequences. Six probability values were calculated for each amino acid in each hemagglutinin. With these six numeric values as independents and the probability of mutation occurrence as dependent, a logistic regression was used to model 753 point mutations from the 241 hemagglutinins to derive population estimates. Thereafter, the future mutation positions in H5N1 hemagglutinin were predicted. Finally, the would-be-mutated amino acids in H5N1 hemagglutinin were predicted by means of translation probabilities between RNA codons and mutated amino acids.