Software merging of trees at best is a "crap shoot". Why? Because unless each tree was produced by the same software, many tags or "facts" are not always equal and information can be lost. Also when the software does a compare of like named people in the matching and merging process you may get false positives that will need to be cleared and for large DBs this could be very tedious.
Andy is also correct about sourcing issues. Many people forget about duplicate sources that may not be merged or the sourcing information may not use the same standards or just many be real bad.
Using a GEDCOM to merge is hard if you are not an epert with GEDCOM. One mistake and your DB is really messed up.