As usual on polling day in Africa, I was standing outside my gate before dawn, awaiting my driver. Paul duly arrived, together with my local assistant, Belinda, and we set off for her nearby suburban polling station. We arrived, with dawn, to find a long queue of keen voters. It is often the case that, unlike in Britain, voters are keen with anything up to 50% of the electorate arriving before the polls open. We passed along the line, indulging in the usual badinage with voters, and entered the polling station to observe the formalities of the opening of the poll - displaying the empty ballot boxes etc. Then I asked those at the front of the queue if they would mind if Belinda voted first so that we could move on to observe other polling stations. As usual, the mention of the magic words "European Union" did the trick and we were able get away quickly.

As the EU's Chief Observer I intended to cover a wide swathe of the Lusaka region, but the next stop put an end to that plan. Paul was registered to vote a Lilanda School, at which there were four separate polling stations serving one of the central Lusaka "compounds". These are areas composed of closely packed small dwellings housing the poorer people of Lusaka, many of whom are unemployed or surviving on pitifully small wages. At Lilanda there were a thousand plus electors either queueing or milling about in the open space between the road and the school. We pushed through the crowd towards the school building, shaking many hands as we went. The polling stations were closed. The polling staff told me that they had no ballot boxes.

With my colleagues, I went off to investigate. With the help of other EU, and also Carter Centre observers, I discovered that sixty-four polling stations, all of them serving the compounds, had no ballot boxes, and that they were currently being delivered by a single truck. Eventually a second truck was found but the delays in commencing voting were still substantial. The final delivery was at 1.30pm - seven and a half hours after the scheduled opening. To their great credit, the Zambian voters remained remarkably peaceful, but these solid citizens, who turned up to vote with great enthusiasm in their Sunday best, deserved better from their electoral commission.

It is hard to believe that the lack of ballot boxes in these particular polling stations was accidental. All the constituencies involved were expected to - and did - vote heavily for the opposition. The immense delays without food or shelter, often till way past midnight, for those who had arrived early in the morning, meant that a number of men and women had to abandon the queue to go to work, to look after children, or simply because they were too old or frail to continue. These were largely votes lost to the opposition.

Three elections were taking place on the same day - presidential, parliamentary and local government - and it became clear that the electoral commission had made no estimate of the time it would take to vote. The slowness of the process, coupled with the lack of extra lines at polling stations with a high number of voters registered, meant that, even without late starts, there were bound to be considerable delays. The polling station at the University, for instance, serving 4207 registered voters, at an average voting rate of 30 per hour, simply could not cope. Eventually a second line was opened but at such a late stage that introducing it caused a number of ballot security problems.

The campaigning period had been marked by continual breaches of the electoral commission's code of conduct. Government vehicles were observed being used by the ruling MMD party; opposition parties were refused permission for meetings if there was any chance of the President arriving in the area; the government owned media - electronic and the written press - continued to be hugely biased towards the ruling party right through the election period; MMD rallies were advertised with the distribution of deeds for houses figuring on the agenda; and, most blatant of all, District Administrators were openly used as MMD party agents, despite being civil servants. In a country with few resources available to the political parties, these abuses were of significant help to the ruling party.

All governments in all countries try it on. Ministers open schools and clinics galore, Humber bridges are promised, and it requires a strong electoral commission or its equivalent to prevent it. Alas, this time in Zambia the commission was part of the problem rather than of the solution. Its chairperson, a High Court judge, openly averred that it was the responsibility of the police and other law enforcement agencies to deal with apparent breaches of the code of conduct. When a police spokesman then directed complainants to the electoral commission it was clear that the ruling party could carry on in its own sweet way with impunity.

By the time the moment to commence the count in the polling station had arrived it was dark and everyone was dog tired, often having been on duty for a full twenty-four hours, and it was hardly surprising that mistakes were made - whether by omission or commission. Once the polling district count was completed the documents and the ballot boxes had to go straight to the tabulation centres where the parliamentary and presidential figures for each constituency were put together for official transmission to the electoral commission.

There is considerable evidence of significant errors in the figures published by the returning officers. For instance, with polling taking place at the same place and same time, with the same electorate, there are twenty-two of the 150 constituencies at which there is a difference in turnout of 900 votes or more between the two polls. Even more striking, in 83 constituencies, for presidential or parliamentary elections, or sometimes both, not a single invalid ballot paper is recorded. If accurate, this would mean that 1,172,529 voters cast their votes without any single one of them making an error. Clearly this is beyond credibility. Such was the lack of independence of the electoral commission that the inauguration of the new President was fixed by the Secretary to the Cabinet and the electoral commission was then bounced into declaring a winner, even before all the results had come in. The announced President had less than 30% of the votes, and a majority of less than 2% over the next candidate. In a situation like this the maladministration and the apparent irregularities assume huge importance.

These points, and others, have been included in the statements of the EU observer mission, for which I and the mission have been severely criticised by the Zambian Government. The domestic monitoring bodies have voiced similar concerns and have come to even more robust conclusions. One respected church body, for instance, stated explicitly that it could not regard the new government as legitimate, whilst another large NGO demanded that the election be rerun. They have not come in for the same attacks. It is clear that the Government's attempts to undermine the EU observer mission are because it is the EU, rather than because of what it has said. The Zambian election has been rather overshadowed by the Zimbabwean situation but, in fact, in terms of electoral tactics, it has many similarities, fortunately without any of the violence.

The problems of democracy in developing countries will not be resolved simply by sending observer missions from Europe and North America. There needs to be a greater commitment to ongoing work with political parties, with the legislatures and with the major NGOs. It is crazy to spend some 12 million euros getting to polling day, as was done by the EU and its member states, with very few plans for post-election projects. Fortunately, the need to integrate projects for democracy building is at last being realised - and not before time.

Michael Meadowcroft was the European Union's Chief Observer at the Zambian elections of 27 December 2001. He writes here in a personal capacity.