Who can say where ideas for novels come from, in what deep slurry of the mind they first take shape? All three of mine had a moment of conception that I can pinpoint exactly: Archangel in the instant of waking from that absurd nightmare, Fatherland in August 1987 when I lay floating in the sea off Sicily listening to the voices of German tourists wafting from the beach, Enigma on the evening of 10 March 1992 when the BBC broadcast a documentary about the mathematician and codebreaker, Alan Turing.

My books, looking back at them, have all suggested themselves as questions. What if Stalin came back to life in modern Russia? What if Hitler had won the war? What if a cryptanalyst at Bletchley Park broke a message he wasn't supposed to read? I suspect this posing of "What if…?" is a fairly common starting point for novelists, especially for those of us who enjoy most the business of telling stories. (I remember once interviewing Stephen King for The Sunday Times. He described how, the previous day, he had watched a child clamber into one of those little mechanical rides you see outside high street shops, and had wondered: "Hmmm. What if he never came out…?")

Having posed the initial "What if…" it is, for me, curiosity as much as anything else which keeps me going during the two or three, or - in the case of Fatherland - five years it takes to turn a passing thought into a 400-page novel. In Archangel, for example, I saw in my mind the two big scenes that occur towards the end of the book - the encounter with the madman in the forest and the train journey to Moscow - almost immediately. It then became a process of working backwards, discovering how my characters could have ended up at this point. I duly read every book I could find about Stalin until, one blessed day, I came across a passage in Dimitri Volkogonov's massive biography, in which he describes how Stalin kept "a black oilskin exercise book in which he would make occasional notes" - a notebook, according to Volkogonov, which disappeared from Stalin's Kremlin safe on the night he suffered his fatal stroke, and which no historian has ever been able to find.

Suddenly I had my story, that odd intermingling of fact and fiction which always sets my imagination running. Suppose this notebook, nearly 50 years after Stalin's death, suddenly surfaced for sale on the black market in modern Moscow. That was entirely plausible, wasn't it? (Just about everything else in Moscow, from sex to SAM missiles, is for sale at the right price.) This immediately, in turn, suggested my central character: a western historian, down on his luck and in need of a big scoop, who'd be prepared to do almost anything to get his hands on such a find. (Plausible again, to anyone who has spent much time with western historians.)

As for the title, I hit on Archangel for no better reasons than that I liked the sound of it, that it summed up the theme of the novel - that Stalin's bloody spirit still hovers above the old USSR - and that it gave my characters a location to head towards. So I went there, too. And this, in many ways, is the part of writing fiction which I most enjoy: standing where I know my characters are going to stand, seeing what they are going to see. Research is too often treated as the poor relation of the literary process: creativity's dutiful drudge. But this, for me, is when a novel begins to come alive. I took the same flight across the endless sub-arctic tundra that my nervous Russian intelligence officer takes. I made the same bone-crunching journey along the unmade road into the forest beyond Archangel ("How long does this track go on for?" I asked my guide. "Two hundred miles," he replied.) I spent the same twenty-two hours on the sleeper back to Moscow, and clambered round the same house in the diplomatic sector (now the Tunisian embassy) where Beria hid the bodies of his victims.

Research, in other words, is not mere hack work which can be contracted out to someone else. You can't write a novel off the top of your head and then go back and add a few quick facts to spice it up. (Or, at any rate, if you do, the results are likely to be disastrous.) The processes of research and creativity are integral. Research gives a book its sense of reality: its truth, its point.

Journalism, I've decided, is both a very good training for a novelist, and a very bad one. It's good because you are professionally accustomed to the idea of getting out of your study and talking your way into seeing things and meeting people. In Moscow, for instance, for the price of a good bottle of malt whisky, I was able to talk to two members of the Russian Intelligence Service (the SVR) whose easy western manners and mild disparagement of their own country helped give me the clues to one of the central characters in Archangel. I've also been lucky enough to see things kept way off the tourist route: Lenin's Kremlin flat, preserved exactly as he left it; the KGB's Black Museum in the Lubyanka.

But journalism is also a very bad habit for a novelist, because at some point you have to throw all this stuff away and simply tell your story. And whereas the essence of good journalism is usually to be as simple and direct with the reader as possible, fiction proceeds best by stealth: by oblique hints and shaded allusion. Facts can be there, of course - in my kind of novels, particularly - but they have to be inherent rather than baldly stated, somehow sieved through the author's imagination. I don't claim to have got this balance right, by any means, but somewhere here, in this tension between fiction and fact - between the sweaty nightmare of the early dawn and the patient pursuit of verisimilitude - lies, at least, my own particular pleasure in writing.