error rate of polymeases

Dr. Duncan Clark news at
Tue Apr 10 05:07:43 EST 2001

In article <200104100326.f3A3Q3814223 at>, the eminent
Wolfgang Schechinger at Academia Sinica wrote
>To my knowledge, the error rate of Taq is said to be 1/1000 , 
>that of proofreading Pols (Pwo, Pfu) ist said to be 1/5000.

Depends on who you quote :-(

Tindall and Kunkel in Biochemistry Vol 27, pp6008-6013 (1988) reported
that Taq produces single base substitution errors at a rate of o for
each 9000 nucleotides polymerised. Frameshift errors were 1 in 41000.

>Are these figures close to reality and what do they actually 

Very dependnat upon reaction conditions in the PCR. Eckert and Kunkel in
NAR, Vol 18, pp3739-3744 (1990) reported up to ten fold lower error
rates if one uses equimolar concentrations of dNTPs and MgCl2.

>When I copy a gene with Taq, 
>there will be (statistically) 1 error in every 1000 bases?

Every one thousand bases polymerised - yes. i.e. potentially two errors
in PCRing 1000bp with one cycle, one per strand.

>Meaning that after a 30 cycle (theoretical) PCR starting from a 
>single copy, there should be 30 errors within 1000 bp (after 
>30 duplications)?

Sounds reasonable but should it be 60? Whatever, theoretically, it is a
lot of errors :-( However we have never seen anything like that many
errors but we tend to use nor more than 15-20 cycles if cloning the
product for expression. Plus we use proof-reading either by a mix or
with say just Pfu. 


Homogeneous Fluorescent Reporting Systems for Real-Time Quantitative PCR: 

Optimisation, Probe Technology & Future Systems

4-5 September 2001
King Alfred's College, Winchester, UK

More information about the Methods mailing list