Filling in the Gaps

Hi all, hopefully someone here can provide a solution to my problem...

Situation: I have over 2 years worth of data grouped by month, the data is the count of the amount errors per different departments in a call center area. (Average of 1200 errors)

Problem: For 2 months (May & June 2007) the data available become corrupted and we were only able to recover around 15% of the current mean (200 errors)

Solution: Is there a way I can implement a formula, process or methodology to fill in these 2 months of data by predicting what the values would be from the rest of the data set I have?
Could people please suggest some solutions!

Thanks!
.