The merge tool will combine two datasets into a single output dataset. The tool contains a 'Field Mapping' option where the user can define which fields from the input datasets will be combined into the output table.
The spatial join tool combines two datasets into a single joined dataset based on a spatial relationship. This tool has the same 'Field Mapping' option as the merge tool. The spatial join tool must be point, lines, or polygons (no tables) since the tool runs on spatial features.
I normally use the merge tool to combine two like datasets (i.e. primary roads and secondary roads) and whose attribute table is similar.
I normally use the spatial join tool to combine two datasets that have a spatial relationship in some way (i.e. traffic signs and roads). Or, this tool can help if I'm needing to analyze attributes from one dataset to another based on a spatial relationship.
Please review the hyperlinks above. Both tools are handy and powerful, but they serve different purposes.
