GCSEs vary "substantially" in their difficulty, with a grade C easier to achieve in some subjects than an E in others, a report suggests.
Latin was found to be the hardest subject to pass
Latin was the hardest in which to gain a C and child development was the easiest, according to the Durham University study.
The difficulty of GCSEs taken by 600,000 pupils in 2004 was analysed.
The exams regulator says it is trying to ensure all subjects are marked according to the same standards.
For the study, by the university's Curriculum, Evaluation and Management (CEM) Centre, the performance of candidates in a specific subject was compared with that in the rest of their GCSEs.
If they did better, it was judged to be easier.
On that basis, it was found to be as difficult to get a grade C in Latin as a grade B in statistics - the "next hardest" subject.
Chemistry, physics and Spanish were third, fourth and fifth most difficult.
The "easiest" subject to get a grade C in was child development. It was found to be as hard as getting an E in Latin.
The next easiest subjects were leisure and tourism and vocational science.
However, the report says there is a need "to be cautious in interpreting these differences as straightforward differences in difficulty".
It cites a "number of reasons", such as pupils being more motivated in certain subjects and schools devoting more time to some than others.
It adds: "The GCSE examination itself may be no harder in that subject, but overall students tend to be less well prepared for it.
"Latin and statistics might be examples of such 'under-timetabled' subjects."
A spokesman for the exams regulator, the Qualifications and Curriculum Authority, said: "This type of analysis is very useful for identifying questions but not for providing answers.
"Many factors could explain why candidates taking some subjects get better results than might be expected and those taking others do less well. How difficult different subject are is only one of them."
He said the QCA was carrying out its own research on "direct comparisons using syllabuses, question papers and the exam scripts of real students, rather than relying on statistical data".