And if your integers are even bigger, there's always [the decimal extension][1]. You can't use indexes anymore, although you could still use a *real* column that approximates your large integers for indexing, and apply the `decimal_cmp` on the subset returned via the *real* index. You'd have to be careful of course, since the *real* approximation could be *above* or *below* the integer, and that complexity is probably necessary only for very large tables, as SQLite will be fast enough even with full-scans in most cases most likely. FWIW. --DD [1]: https://www.sqlite.org/floatingpoint.html#the_decimal_c_extension