The accuracy of the last census has been called further into question with the publication of a damning report by the government's own statistics watchdog.
The Statistics Commission said there was 'too much uncertainty' about the results in central London for full confidence to be placed in the survey.
The Office of National Statistics must do more to explain its methods, said the commission, and look at alternatives to the 'usually resident' measure of population.
Commission chairman David Rhind said the results of the census in Westminster were 'substantially less reliable' than the margins of error applied would suggest.
'When the initial enumeration misses as much of the resident population as it did in Westminster, even sophisticated estimation techniques may not entirely compensate for the initial absence of reliable data,' he said.
'In simple terms, there is too much uncertainty about the final results in the most 'hard to count' areas.'
He added: 'The Office for National Statistics did not have enough reliable information available to provide a confident estimate in this case. We believe the problems of estimation were compounded by the fact the concept of a 'usually resident population' is particularly difficult to measure in the centre of one of the world's major cities.'
Kit Malthouse (Con), deputy leader of Westminster City Council, said the report compounded pressure on the government to look at the census again.
'With the publication of this report, it is time for the ONS to stop defending their mistakes and start correcting them. This saga again calls into question the ONS's position as official statisticians to the government, especially when the funding of vital services to millions of people depends on them getting their sums right,' he said.
The ONS said it had been conside ring a recount in Westminster but that no decision would be taken until the ongoing data matching exercise is completed, which is expected to be in the New Year.