Yes, there is and the rate varies according to the sequencing technology used: A really frustrating aspect of modern ‘long’ read technology is how ‘noisy’ the sequencing results are. A frequently used rule of thumb is that a sequence read with less than one error per 10,000 nucleotide base-pairs is acceptable. This quality is often expressed on the ‘Phred’ scale, originally used for a program written by Phil Green to ‘call’ bases on traces obtained from a DNA sequencer made by a company called ABI. A Phred score of Q = 40 represents an error rate of 1 incorrect base in 10,000 bases read. The formula to calculate Q is:
Q = -log10(P)
Where P is the probability of an incorrect base call = 1 / 10,000 in this example.
Scottish Wildcat nailed it, well done.
So to sum up, there is a certain error rate with DNA sequencing, as with every measurement you conduct, be that in biology, chemistry, physics or other disciplines.
They way we deal with that in genome sequencing is that we sequence every basepair several times independently. That means that we can identify and correct sequencing errors.