What is the most apt image of literary influence? According to Borges we invent our precursors, while TS Eliot had it that great poets are thieves, not imitators. If Harold Bloom is correct  and he's been quite sure of himself for almost 40 years  the placid scene of influence is in reality a brawl, with writers engaged in pugilistic agon against their aesthetic progenitors. The great merit of Bloom's 1973 book The Anxiety of Influence was to have turned a weak critical term  the word itself having declined from its Shakespearean sense of inspiration  into a call to arms. For a time, the unclubbable Bloom ("I am a department of one") was ranked with his Yale colleagues Paul de Man and Geoffrey Hartman as an American critical iconoclast to match those visiting-prof affronters, Barthes and Derrida.

Reading The Anxiety of Influence as a student in the late 80s was a revelation; the seamless begetting of literary tradition suddenly looked like the most rancorous of Freudian family rows. But already (and especially in light of his slightly comical notion of the "strong poet") one suspected Bloom of being, or having become, another sort of critic: ruinously addicted to ex cathedra pronouncements, and a deal less given to actual argument than you'd hope for in a writer whose hobbyhorse was the intricate drama of doubt and persuasion itself. The Anatomy of Influence has some of the strengths and all of the faults of Bloom's subsequent work.

The book, he tells us, is his "final reflection on the influence process". He is now in his 80s and, following a number of health scares and other late-life catastrophes, finds himself in melancholic mood, planning a capacious, personal and baroquely styled critical study of some of his own abiding influences: Shakespeare, Yeats, Whitman, Emerson and Hart Crane among them. There is much to admire and regret in the essays that follow, but first one has to negotiate a "Praeludium" (a mere preface would not do for Bloom) of amazing and almost forgivable egotism. It's here that he rails for the umpteenth time against lesser critics who have dared to suggest that his boisterous, agonistic account of writerly influence might be weighted in favour of a certain masculinist tradition. Here too that the usual Bloomian oracular tone is established in fathomless aperçus: "Literature for me is not merely the best part of life; it is itself the form of life, which has no other form." This before we have got to his oddly repeated boasts about how much poetry he can still recite by heart.

None of this exactly bodes well for the essays on Shakespeare with which the book properly begins. Here again is the grandstanding more suited to callow junior fellows: "When I was in Denmark to receive the Hans Christian Andersen bicentennial award . . ." But once Bloom has got past this and the gnomic clichés  "Hamlet contains virtually all women and men"  his readings of Shakespeare, and his influence, are at least engaged and sometimes daring. The insight that for all their eloquence and generous humanity Shakespeare's characters never really listen to each other is a fine one, and deserves to be elaborated, though Bloom would never be patient enough to do it. He is suggestive too on the Shakespeare of Ulysses and Finnegans Wake  "Joyce's defence against Shakespeare is total appropriation." And although crudely expressed, his scorn for strictly historicist readings of the plays is principled and probably right.

The essays on poetry are equally mixed. Bloom has long had a woeful love of the lofty but limiting epithet, so that here Crane is "the Orphic Hart Crane" and Yeats "the occult seer Yeats". Too often, his narratives of literary inheritance devolve into biblical lists, the specificity of influence quite lost: "Robert Browning was Shelley's principal heir as Tennyson was Keats's. After that, the sequence is remarkable: Swinburne, Shaw, Yeats, Hardy, Forster, Woolf, and, surprisingly, Joyce and Beckett. Shelley fuses with Hardy and Whitman in DH Lawrence's poetry." (One is put in mind of Pete Frame's Rock Family Trees diagrams  every member and ex-member of Hawkwind, plus associates.) And yet, he can suddenly turn in a bravura reading of a poet such as Whitman, making his homoeroticism seem to turn on a single word, in this case "tally". It's in the pieces on American writers that the critic's own literary self-fashioning starts to cohere: Bloom's great influence is not Shakespeare or Yeats, but Emerson. He is in his own last analysis an individualist, albeit (like the writer who saved him, he says, from a mid-life crisis) a self-contradictory one.

The odd corollary of all that infuriates in this book  the pomposity, the lists, the lack of close reading  is that one suspects Bloom of still being a great teacher. In a telling aside, he says he teaches only undergraduates; most venerable academics of his reputation, let alone age, would have given up long ago. Bloom has always been an admirer of an aphoristic aesthete tradition alongside his ostensibly more robust canon of poets. He has reached an age and standing that allow for the rudest of one-liners, as when he calls Matthew Arnold "the most overrated of all critics, ever"  it's the petulant "ever" that truly endears  or laments "the egregious Edgar Allan Poe". If he goads a few nascent academic critics into similar facetiousness, his influence might not be squandered after all.

Brian Dillon's Sanctuary is published by Sternberg Press.