Counting the number of malaria cases worldwide is more guesswork than science.
Estimates vary from 300-500 million attacks a year, with the number of deaths at between one and three million.
At least 90 per cent of these cases occur in Africa south of the Sahara and here lies the challenge.
On the one hand, many episodes of fever, particularly at primary health care level, are diagnosed as malaria.
This trend is encouraged by international agencies like the World Health Organization which would rather over-treat fever than see an infant die through delayed treatment.
On the other hand, many cases of malaria simply go unreported. Families purchase drugs at shops or pharmacies and cure the illness at home.
Professor Zul Premji, head of the department of parasitic infections at the Muhimbili University College of Health Sciences in Dar es Salaam, Tanzania, is among those who think that the figures are inflated.
"In 1999, official data announced by the Ministry of Health confirmed 16 million episodes of malaria in Tanzania that year.
"That is actually the data for all febrile illnesses. Because the diagnostic facilities are lacking, we still rely on fever as the main indicator of malaria."
Premji says his own research reveals that in endemic areas, children aged between five and 24 months have on average four to five episodes of malaria a year.
"As the child grows up and acquires immunity, the number of episodes do decrease, and by the time the child is ten to 15 years, it is usually one episode of malaria a year. That is the sort of burden we are talking about."
In contrast, Canadian epidemiologist, Don deSavigny, points out that malaria contributes significantly to other child deaths.
"Most children in endemic areas where malaria is common are getting repeated episodes of infection and illness. You end up with a very vulnerable child.
"The next illness they acquire might be diarrhoea or pneumonia which a child would normally be able to fight off, but after three or four episodes of malaria, they will die from it."
"That statistic then becomes a malnutrition death or a pneumonia death or a diarrhoea death, but underlying it was the malaria."
DeSavigny points out that successful malaria control measures have led to a far larger drop in the number of deaths than could possibly be due to malaria alone.
"When we prevent malaria, we expect to reduce mortality by at least ten per cent but in fact when we use an insecticide treated bed net, the mortality reductions are much, much larger.
"In some places, we reduce mortality by 20, 30, 40 per cent and we reduce episodes of illness by over 50 per cent. This suggests we have been underestimating the role of malaria as a burden."
The debate is more than academic. Without accurate data it is hard to target resources effectively, especially in a country like Tanzania, which has one of the highest burdens of disease in the world and where annual public health expenditure works out at about $3 a person.
In 1996, when the Kinet project using insecticide treated mosquito nets was launched, a demographic surveillance system (DSS) was set up in 25 villages in the Kilombero and Ulanga districts of southeastern Tanzania.
The goal was to gather accurate information on a rural population of about 65,000 people.
The districts lie in the flood plain of the Kilombero river. The roads are unpaved and during the rainy season, some villages are difficult to reach for several months each year.
Villagers are generally subsistence farmers growing rice and maize. About 40 per cent of the population are Muslim and 60 per cent Christian.
Most of the houses are made of thatched roofs and mud walls. Wealth here is a tin roof, a bicycle, the possession of a radio.
Until the DSS was up and running, little was really known about births, deaths and illnesses at the household level.
Leave regions, districts, divisions and wards behind. Exit the larger villages and you will find yourself in hamlets or kitongoji - Tanzania's smallest administrative units.
Here, if you head off the beaten track, you might stumble across characters like Hadija Nyanga, pushing her bike through a paddy field or wading across a flooded marsh, clipboard in hand, doing the spadework or rather the legwork on which the DSS depends.
Although houses are mainly concentrated along the main roads, farming or shamba huts where the family will spend many weeks each year can be several hours' walk away and are widely scattered.
Depending on where she can find a senior family member, Hadija does her work at either the shamba house or the village house, asking about births and deaths, pregnancy and sickness, migration and employment.
Looking for a farmer in a flood plain is slightly easier than looking for a needle in a haystack but still immensely time consuming.
Hadija reckons on visiting about ten households a day. She will return to a household every four months to update the information.
To be able to trace people, every village has a code and each household and individual has been assigned a unique number. Coordinate all three and you have a reliable means of identification.
Hadija Nyanga is one of about 27 field interviewers. Field supervisors oversee their work and help resolve any inconsistencies. By an elaborate system of checks and controls, the DSS hopes to gather accurate data.
Regional statistics are later extrapolated from this information, which is the basis upon which health care policies are formulated.
Five donors are currently supporting the DSS - the World Health Organization, the Swiss Tropical Institute, the Swiss National Science Foundation, the Swiss Development Agency and USAID.
swissinfo, Vincent Landon
Malaria probably kills about 80,000 people in Tanzania each year.
These are mainly children under five and pregnant women.
It is responsible for a quarter of all child mortalities and 20 per cent of all maternal deaths.
It is the major cause of illness with an estimated 16 million cases annually.
About 35 million people live in Tanzania.
Malaria consumes $119 million of state and household expenditure a year.
In terms of national resources, that's about 3.4 per cent of GDP.
In compliance with the JTI standards