There's a use case when STR() works better/is easier than other methods of converting numbers to strings - when the numbers are already typed as float or real. CAST and CONVERT have some limitations on converting float and real types to strings (scroll down to Remarks and see the second subheading, "float and real Styles").

I recently wrote about a real-world use case where STR() avoided a problem caused by these limitations of CAST and CONVERT with float and real values.[/url] While the problem could be solved by converting the float/real values to decimal values and then converting them to strings, the STR() function does the same thing without having to code the nested conversions.