depends on type - those shown as extracted records are generally very good though I have come across a couple of typographical errors eg one was 100 years out - the main error is the confusion between old and new style dates and that they did not always understand latin (eg eod diem or similar = same date for successive entries) thus entries pre 1753 in jan thru mar always need special checking
.
Those marked patron submitted fall into three classes
a - those done as a batch replacement for extracted records (eg most of Michael) are generally not too bad but they do include multiple entries (different spellings) for a single entry - one batch for St Marks is however best described as nonsense

(b) those submitted privately - generally a one family or one name study - these range from the accurate to the totally fictional - you cannot tell - names of wives are often added that are not in the original record

(c) one-offs - often totally weird mis-readings from off Island - many early records and/or strange churches are in this category - junk - anything in 'sealings for the dead' should also be treated as fictional.

the only way to tell is to borrow the film of the 1911 official transcript and check that (that is the basis of the extracts on my CD) - I have found this 1911 to be very very accurate but often conservative in that if you can see the original register you may be able to make out some more of a damaged record - the other place for records post 1733 are Bishops transcripts which are supposed to be an annual contempory copy of the register but not all have survived -on film or actual record at Manx Museum.

The Burial reg as transcribed by the IoMFHS and basis of Brian's index has quite a number of date errors usually old/new style dates eg the patrick register the first entry explicity states it was the first burial in the yard but the IoMFHS has it a year later because they confused the years. There is also a significant error rate especially in some parishes with pages missed out - Brian's index is a transcription of this transcription with an additional error rate rangeing from 2-3% to over 8% (jurby) in those parishes and periods I have checked - again borrowing the parish reg film allows you to check those pre 1880'ish