A surrogate standard usually has similar chemistry to the analyte of interest and is added to the raw sample (before any processing). Its role is to investigate the effect of sample processing methods on the analyte of interest. For example you can use it to investigate recovery in an extraction procedure.
An Internal standard (I.S) on the other hand is added just before analysis. It helps to take care of variations in signal due to the instrument...The rule is "what happens to the "analyte" will also happen to the I.S. Therefore when you plot your calibration curve using the analyte:I.S ratio you get more accurate results
Both of these answers are quite good. A more important question might be why this is done. This is a common practice in envioronmental applications where the recovery of the analyte is suspect or expected to be low. A surrogate standard is added to the sample prior to cleanup or processing, the thought being that this can be used as a gauge as to the analyte recovery. The IS is added to the sample after the cleanup process so as to perform quantitation.
In my opinion, there is no great difference between a surrogate and an internal standard. Let's think: a surrogate (or IS) should always be added in the beginning of the analytical process (preferentially at the sample collection). If the IS (or surrogate) was well chosen (i.e. very similar physicochemical characteristics compared to the analyte, but with analytical signals differentiable by the instrument), it is expected that the same matrix effects and recoveries would be experienced for both analytes and IS (or surrogate). As a known amout of IS is added in the beginning and the analytical signal used is the analyte Peak area/IS peak area, I cannot see any reason why one should add another substance just before instrumental analysis. Unfortunately, this concept is not very well understood for some researchers, and the IS is only added just before analysis, and the results obtained are used to calculate analyte recoveries. Well, by doing so, analyte recovery is not adequately evaluated and erroneous results may occurr.