Which is the correct way to statistical yzed multiple exponential correlations?

I need to yze for measurement accuracy with statistic tools, four sets of (x,y) data. Its the same measurement (change in concentration v. time). The correlation that best matches the information is a exponential function. Can you point me to the correct procedure? also, how can you "eliminate" a point that obviously is wrong and ruining the information (using statistics tool also)