Erlang PostgreSQL Database Client Pool for epgsql Configure application env like so: application:start(epgsql), application:set_env(epgsql_connpool, pools, [ {maindb, [ {host, "127.0.0.1"}, {username, "postgres"}, {password, ""}, {size, 10}, {opts, [ {timeout, 5000}, {database, "my_database"} ]} ] } ] ), application:start(epgsql_connpool). And to make queries: {atomic, NumRows} = epgsql_connpool:transaction( maindb, fun(_Db) -> Sql = "SELECT count(*) FROM my_table WHERE id = $1;" {ok, _, [{Ret}]} = epgsql_query:equery(Sql, [123]), Ret end). With transaction(), all queries made within the fun are surrounded by a BEGIN/COMMIT/ROLLBACK. The fun passed to epgsql_connpool:transaction/* will be handed a pid as its first argument. epgsql_connpool stores the connection pid in the process dictionary. By storing the pid, calls to the epgsql_query module (which is very similar in interface to the pgsql module) don't need to pass the pid directly. To support older applications, the pid can either be used to make calls to the pgsql module directly. With dirty(), the statements are run as-is, not in a transaction.