Paulo G., that was a great catch!

BETWEEN didn't even cross my mind when looking at this originally. I was looking more at the generalities of which techniques worked best with a multi-value filter and totally missed the fact that my filter data were sequential.

If we change our sample data to something like (8, 214, 271, 310), NOT EXISTS keeps its performance edge, but the best solution should take into account the filter data, and the BETWEEN solution crushed the NOT EXISTS solution for the given sample (214-217).

Thanks for the new angle!

KenJ