The performance characteristics of LC in three types of narrow bore columns are compared: (A) small bore packed columns, (B) packed capillary columns, and (C) open tubular columns. The available experimental data confirm that reduced plate height/reduced velocity curves (h, ν curves) for types A and B are similar to those for conventional wide-bore packed columns, that the permeability of type A columns is the same as conventional packed columns, but that type B columns are about three times more permeable. Experimental data for open tubular columns confirm that with 50 μm bore tubes the Golay equation gives a good approximation to actual behaviour. Theoretical analysis indicates that while it is possible to operate type A columns of 10,000 plates upwards under optimum conditions with an injector/detector having an effective volume of 0.1 mm3, there are no practical circumstances under which type B and C columns can be so operated. It is further shown that even though type B columns are three times more permeable than type A columns, it is still not possible with feasible injector/detector systems for them to compete In terms of performance with type A columns. Open tubular columns, on the other hand, because of their 100 times lower separation impedance under optimum conditions, can compete with type A columns if a very large number of plates is required. Foradetector volume of 0.1 mm3 the cross-over occurs at N ∼ 500,000 with a 27.5 μ m bore column 120 m in length and a time for the unretained peak of about 8 hr. Consideration of the optimum ratio of stationary to mobile zone indicates that the wall layer in an open tubular column for LC should be about 25% of the diameter of the bore, giving a stationary to mobile zone ratio of about 55 to 45.