Off of the same wafer that's probably about right.
The amazing thing is the stated limits on Vbe, when shown in datasheets, compared to the typically observed ones. For example the BC847 says min. 580mV, typ. 660mV, and max. 700mV, at a given opertaing point (5V Vce, Ic 2mA). For a well-controlled process and consistent starting materials these limits are probably at the >6 sigma level.
Try to sort out of parts with the same date code, although this is not a guarantee everything is from the same wafer. Also just because parts are on on a tape doesn't mean they are the same batch---see my data from 1N4148's in the expensive limiter thread, where there were clearly two populations in a batch.
There are now some manufacturers (Ph*lips, D*odes Inc. are two) who have got a system working where adjacent die are plucked off the wafer and paired. The matching is almost as good as monolithic pairs but is not tested typically iirc. They are pretty cheap. The higher isolation between the two devices is an advantage in some cases. I tried to get company D to tell me what the breakdown voltage between devices was and it hadn't occurred to them to find out---initially the guy said "infinite." :razz:
If they could make some quads that would be nice.