diff --git a/src/mos6502.addr.c b/src/mos6502.addr.c
index dd63c09..ba8bae7 100644
--- a/src/mos6502.addr.c
+++ b/src/mos6502.addr.c
@@ -12,6 +12,11 @@
 #include "mos6502.h"
 #include "mos6502.enums.h"
 
+/*
+ * This is a table of all the possible opcodes the 6502 understands,
+ * mapped to the correct address mode. (Well -- I _hope_ it's the
+ * correct address mode!)
+ */
 static int addr_modes[] = {
 //   00   01   02   03   04   05   06   07   08   09   0A   0B   0C   0D   0E   0F
     IMP, IDX, NOA, NOA, NOA, ZPG, ZPG, NOA, IMP, IMM, ACC, NOA, NOA, ABS, ABS, NOA, // 0x
@@ -39,9 +44,9 @@ static int addr_modes[] = {
  * cpu) that an instruction will use.
  */
 mos6502_address_resolver
-mos6502_get_address_resolver(int addr_mode)
+mos6502_get_address_resolver(vm_8bit opcode)
 {
-    switch (addr_mode) {
+    switch (mos6502_addr_mode(opcode)) {
         case ACC: return mos6502_resolve_acc;
         case ABS: return mos6502_resolve_abs;
         case ABX: return mos6502_resolve_abx;
diff --git a/src/mos6502.c b/src/mos6502.c
index 649b622..c8fcc57 100644
--- a/src/mos6502.c
+++ b/src/mos6502.c
@@ -309,3 +309,47 @@ mos6502_get_instruction_handler(vm_8bit opcode)
 {
     return instruction_handlers[mos6502_instruction(opcode)];
 }
+
+/*
+ * This code does the execution step that the 6502 processor would take,
+ * from soup to nuts.
+ */
+void
+mos6502_execute(mos6502 *cpu, vm_8bit opcode)
+{
+    vm_8bit operand;
+    int cycles;
+    mos6502_address_resolver resolver;
+    mos6502_instruction_handler handler;
+
+    // First, we need to know how to resolve our effective address and
+    // how to execute anything.
+    resolver = mos6502_get_address_resolver(opcode);
+    handler = mos6502_get_instruction_handler(opcode);
+
+    // The operand is the effective operand, the value that the
+    // instruction handler cares about (if it cares about any such
+    // value). For example, the operand could be the literal value that
+    // you pass into an instruction via immediate mode. As a
+    // side-effect, resolver will set the last_addr field in cpu to the
+    // effective address where the operand can be found in memory, or
+    // zero if that does not apply (such as in immediate mode).
+    operand = resolver(cpu);
+
+    // Here's where the magic happens. Whatever the instruction does, it
+    // happens in the handler function.
+    handler(cpu, operand);
+
+    // This will be the number of cycles we should spend on the
+    // instruction. Of course, we can execute instructions pretty
+    // quickly in a modern architecture, but a lot of code was written
+    // with the idea that certain instructions -- in certain address
+    // modes -- were more expensive than others, and you want those
+    // programs to feel faster or slower in relation to that.
+    cycles = mos6502_cycles(cpu, opcode);
+
+    // FIXME: actually emulate the cycles
+
+    // Ok -- we're done! This wasn't so hard, was it?
+    return;
+}