This paper highlights the discrepancies found in design codes for very long externally pressurized cylindrical shells subjected to uniform external pressure: although the failure is identical to the column buckling of a ring subjected to an external compressive line load, design codes provide far smaller design values of the buckling load for the cylindrical shell in comparison to the ring. Based on the Eurocode 3 (EN 1993-1-6) on shell buckling this situation is pointed out and validated by numerical studies. The results are interpreted and compared to findings from experimental studies for providing an adopted design proposal for very long cylindrical shells subjected to external pressure. The numerical findings are verified by a set of experimental tests with a series of small cylindrical shells subjected to external pressure. The tests provide the fundamental basis for the proposed improved assessment procedure for cylindrical shells.