Drug related deaths often attract a great deal of political and media attention. Statistics on drug related deaths, along with anecdotes from individual cases, are often used to illustrate the harm drugs can cause and to defend prohibitionist policies.
Although these statistics can be used to provide one indication about how harmful drugs are, it is an overly simplistic indicator for this purpose. Most drug related harm is non-fatal, and drug related deaths are not proof that a certain drug is inherently harmful, since harm to an individual is determined by how the drug used (dose, frequency, poly drug use, method of use, etc)
Illegal drugs are often sold in unknown strength and purities, and in preparations that maximise the dealer's profits rather than protect the health of the user. This will make some drugs much more harmful then they might otherwise be if they were sold in regulated environment.
Many drug users use more than one drug (including legal drugs such as alcohol) which makes identifying which drug(s) were responsible for the cause of death tricky.
The Office for National Statistics (ONS) produces mortality statistics for drug related deaths based upon information on death certificates. All deaths that mention drugs on the death certificate are counted, which includes deliberate poisonings (ie homicides and suicides).
In 2000 the Advisory Committee on the Misuse of Drugs published a report which included an extensive critique of the ONS statistics. It described the system for generating the data as "flawed" and said that "the ONS approach to reported drug-related deaths was unsatisfactory for our purposes in that it will also capture cases other than drug-misuse related deaths".
In 2001 the ONS introduced a new definition based upon ACMD recommendations. This methodology still has a number of serious shortcomings which the ONS acknowledge in their statistical bulletins :
Like much of the Government's drug statistics, two of the most commonly used drugs – alcohol and tobacco – are not included in the same datasets. We have collected figures for these drugs to provide some comparison, but due to different methodologies used for collected the data, caution must used when making comparisons between alcohol and tobacco and the other drugs.
The ONS produces data for alcohol related deaths but it is a different (and broader) definition than used for other drugs as it only includes deaths regarded as being most directly due to alcohol consumption. This excludes homicides and suicides but does include deaths for chronic liver disease and cirrhosis, even when alcohol is not specifically mentioned on the death certificate.
Smoking history is rarely recorded on death certificates and the ONS does not produce figures for tobacco related deaths. The Health Development Agency has produces a report which includes an estimate for the number of smoking related deaths based upon the contribution that smoking makes to specific conditions recorded at death. 
* The ONS's most recent stats on alcohol related deaths only include figures for whole of the UK. We have contacted the ONS and they have provided us with figures for England, Wales and Northern Ireland.
Alcohol-related deaths in England, Wales, and Northern Ireland (2004-2007)
Alcohol related deaths are measured using a broader definition compared to other drugs. See commentary above for more details.
Following recommendations from the Advisory Council on the Misuse of Drugs, both the Office for National Statistics and the General Register Office for Scotland now use the same definition for drug-related deaths.
We have included data for alcohol and tobacco for a general comparison, but care must be taken when comparing it with other drugs as the methodology used is not the same [see England and Wales for more information].
Note: the General Register Office for Scotland's drug related deaths for 2006 appear to indicate that there were 131 deaths in Scotland involving alcohol. However this figure appears to only refer to deaths where alcohol was preset in addition to an illegal drugs and therefore does not include deaths where alchol was the only drug involved.