• We experienced the exact same issue generated by both the reference of temp tables and a statement that includes OPTION(RECOMPILE) for any statement executed after the point in time the sp_recompile is issued.

    This is obviously a problem when a statement level recompile occurrs and there is an in flight execution of the proc when sp_recompile is issued.

    In a high concurrency environment it makes issuing sp_recompile to overcome runtime stats / sniffing issues a bit of a gamble, particulary if issued against a table with many dependant procs.