Introduction and caveat

This question is virtually unanswerable, firstly because there is very little good data. Other than the United States, there have been very few demographic studies asking such a question in other countries. Furthermore, studying this subject is extremely tricky for 4 big reasons:

1. The problem of definition. Much hinges on how the researcher defines "gay", and how his subjects perceive his questions. Is the definition based on sexual activity, if so, how frequent must a subject engage in same-sex activity, before he is classified as gay? And what does "sexual activity" or "sexual relations" mean? Different people have quite different ideas of what constitutes sex, and surveys often trip over this issue. Some people have sexual relations exclusively with same-sex or opposite-sex partners, but many, in different phases of their lives, may swing various ways. How does one classify these subjects? Some surveys have asked respondents if they identified as "gay". But many people who engage in same-sex activity, or who feel attracted to persons of the same sex, do not identify as gay. Many others may be attracted to the same sex, but for lack of opportunity or whatever reason, do not act on it.
2. Sampling techniques vary greatly from one study to another. Many surveys set out to study something other than the demographic percentage, e.g. household patterns, or risk of sexually transmitted diseases, and the sampling method, while suitable for the main purpose of the study, may be unsuitable to drawing conclusions about the percentage of homosexual persons in the general population.
3. Sex is a highly personal subject, and answers obtained are very sensitive to the survey methods used. One such study, often cited by anti-gay organizations, is that by the Alan Guttmacher Institute, entitled The Sexual Behavior of Men in the United States, released in 1992. It reported that only 1 percent of its participants identified themselves as exclusively gay, but its method was to approach respondents door to door in mainly suburban areas!
4. The problem of social context. Homosexuality is stigmatized, censured by many religious leaders, and criminalized in some countries. People may fear for their jobs, or their place in their families if anyone should discover they are gay. Under these circumstances, there is a high likelihood of substantial under-reporting of the incidence of homosexuality in any survey. Even if the survey method is impersonal and a researcher promises anonymity, what benefit does a participant get for trusting him? What motivation is there to tell the truth compared to the strong reasons to lie?

These above-mentioned difficulties must be borne in mind when reading further about demographic studies. Below are some useful URLs.

 
But what's your motive?

What is the purpose of asking how many people in a population is gay? Certainly, it is a minority, however one defines it, but does it matter whether it is x% or y%?

Is respect for a minority, their liberties and rights, to be calibrated based on numbers? Do people become less equal in law and in our conscience, the fewer they are?

The Sikh community in Singapore numbers less than 1% of our population. We make exceptions to our motorcycle helmet laws and military uniform regulations, in deference to their customs and beliefs. When the Sikh community celebrated the opening of their new Central mosque in 2002, the Prime Minister officiated, and wore a turban as a sign of respect.

 
Useful URLs

(note: a simple websearch will yield many websites discussing this subject that are maintained by anti-gay organisations, mostly of a US Christian persuasion. Their purpose is often to highlight the lower range of figures, to dispute the higher figures and to suggest that homosexual persons are too few to be accorded equal rights.)

 
San Francisco Public Library - gay demography

This webpage of the San Francisco Public Library gives a brief overview of the difficulties of interpretation of data, and a summary of some important studies.

Quotes:

The 1993 Janus Report, the first broad-scale scientific national survey on sexual behavior since Kinsey, concluded that 9% of males and 5% of females had had homosexual experiences more than just "occasionally."

The 1993 Yankelovich Monitor Survey, considered the first nationally representative survey to reflect what percentage of the population identified itself as homosexual, indicated that 5.7% described themselves as "gay/homosexual/lesbian."

The 1994 Sex in America Study Self-identified gay and bisexual men accounted for 2.8% of the surveyed respondents, while 1.4% of the women identified as lesbian or bisexual.

 
Kinsey Institute - Prevalence of homosexuality (1999)

This site gives a summary of various demographic studies to 1999. Smith (1991) classified 5-6% of adults as homosexual or bisexual since age 18. Taylor (1993) found more than 4% of men aged 16-50 and more than 3% of women in the same age group reporting a same-sex sexual partner in the previous five years.

Laumann et al (1994) found that 9% of men and 4% of women reported having engaged in at least one same-gender sexual activity since puberty. Given the identity category choices of heterosexual, homosexual, bisexual, or something else, 2.8% of men and 1.4% of women surveyed reported "some level of homosexual identity." 

Binson et al (1995) showed 5.3% of men reporting sexual activity with a same-gender partner since age 18. and 6.5% of men reporting sex with men during the previous five years. The highest prevalence was found  the 12 largest cities (14.4% since age 18) and among "highly educated" White males (10.8%).

Sells et al (1995) reported 6.2% of  males and 3.6% of  females with "sexual contact with someone of the same sex only or with both sexes in the previous five years," and 20.8% of U.S. males and 17.8% of U.S. females with some homosexual behavior or some homosexual attraction since age 15. 

Gonsierek et al (1995) reviewed the literature and critiqued surveys of homosexual activity from Kinsey in 1948 to the 1994 study by Laumann, et al. Because of the possible risks involved in self-disclosure, it is posited that the recurrent 2-5% for same-gender sexual behavior in the studies reviewed represents a minimum figure. They suggest that the current prevalence of predominant same-sex orientation is 4-17%.

 
Virtual City - Calgary study

This webpage reports the findings of a study by Bagley and Tremblay of 750 males Calgary, a small city in Canada, aged 18 to 27. They found that 14% had had male-male sex, 5.9% identified as homosexual and 6.1% identified as bisexual.

 
60 years of counting queers - Defense Research Institute

This webpage summarises the findings of a report by RAND's National Defense Research Institute, for the US Dept of Defense. This study did not involve new research, but an analysis of those studies which, in the opinion of the analysts, conformed to high methodological standards. These included:

QUOTE:

• National Opinion Research Council, 1970:
  Male 6.7 percent
  (Lesbian n/a)
• General Social Survey:
  Male 5 percent
  Female 3.5 percent
• Louis Harris & Associates, 1988:
  Male 4.4 percent
  Female 3.6 percent
• Research Triangle, 1991:
  Male 8.1 percent
  (lesbian n/a)
• National Survey of Men, 1993:
  Ages 20-39, 2.3 percent

 
60 years of counting queers - National health and Social Life Survey

This webpage gives a summary of the National Health and Social Life Survey conducted in 1992 by the National Opinion Research Center (University of Chicago). Based on a random sample of 3,432 U.S. residents, it found that 2.8% of men and 1.4% of women self-identified as gay. The study also found that in the 12 largest metropolitan areas, gay-identified men were 10.2% higher than in the suburbs, (corresponding figure for women was 2.1% higher).

 
60 years of counting queers - Guttmacher Institute survey

This webpage discusses the study entitled The Sexual Behaviour of Men in the United States, released by the Alan Guttmacher Institute in 1992. It received much media attention because it only 1% of its participants identified as gay, well below other studies.

QUOTE:

This conclusion was not surprising since most of the respondents were approached door to door in mainly suburban areas of the United States, where gay and lesbian Americans are least likely to disclose their sexual identities.

UNQUOTE

 
Tasmanian Dept of Education

This webpage is from the Tasmanian Department of Education. It includes an interesting paragraph:

QUOTE:

How many students are affected by sexuality discrimination and harassment? 11% of Grade 7 to 12 students surveyed by La Trobe University academic Lynne Hillier for a study of student sexual health did not identify as heterosexual. The study of nine schools around Australia included three in Tasmania. Other recent national student sexual health surveys have found similar percentages. A recent study of homophobic attitudes at Hobart's Elizabeth College found that 9% of students did not identify as heterosexual.

UNQUOTE

 
Sex in Australia survey

This webpage mentions a study by Smith A, Rissel C, Richters J, et al. Sex in Australia: sexual identity, sexual attraction and sexual experience among a representative sample of adults. Aust N Z J Public Health 2003; 27: 138-145.

QUOTE:

A recent Australian telephone survey that included 9134 women aged between 16 and 59 years, randomly selected from all states and territories, provided a wide range of information regarding sexuality. While 0.8% of the women identified as gay and 1.4% as bisexual, 15.1% reported same-sex attraction or sexual experience. It is reported that 8%–11% of young people have a non-heterosexual orientation.

UNQUOTE

See also the abstract of this study. It was a telephone survey, asking people intimate details about their sexual identity and sexual experiences!