The open source Tamarin project includes a reference table for the opcodes used in the ABC format.

See SWF9Tools for some tools that can help with the reverse engineering of the opcodes.

See SWF9DIS for the main page of this project.

See http://labs.macromedia.com/wiki/index.php/ActionScript_3 and http://livedocs.macromedia.com/labs/1/flex/langref/index.html for more information about ActionScript 3.

Will invent names of opcodes similar to those of MSIL. See ECMA-335: CLI Partition III - CIL on http://msdn.microsoft.com/netframework/ecma/ for documentation on MSIL instruction set.

03 : throw

04 : getsuper

05 : setsuper

08 XX : regreset ??

09 : nop

0C XX YY ZZ : bge - branch on greater than or equal. ZZYYXX is offset in bytes.

0D XX YY ZZ : bgt - branch on greater than. ZZYYXX is offset in bytes.

0E XX YY ZZ : ble - branch on less than or equal. ZZYYXX is offset in bytes.

0F XX YY ZZ : blt - branch on less than. ZZYYXX is offset in bytes.

10 XX YY ZZ : br - unconditional branch. ZZYYXX is offset in bytes.

11 XX YY ZZ : brtrue - branch on true. ZZYYXX is offset in bytes.

12 XX YY ZZ : brfalse - branch on false. ZZYYXX is offset in bytes. Seen with logical-or ||.

13 XX YY ZZ : beq - branch on equal. ZZYYXX is offset in bytes.

14 XX YY ZZ : bne - branch on not equal. ZZYYXX is offset in bytes.

15 XX YY ZZ : bnge - branch on not greater than or equal

16 XX YY ZZ : bngt - branch on not greater than

17 XX YY ZZ : bnle - branch on not less than or equal

18 XX YY ZZ : bnlt - branch on not less than

19 XX YY ZZ : bseq - branch on strict equal

1A XX YY ZZ : bsne - branch on strict not equal

1B : switch

1C : xmlop3

1D : popscope

1E : forin

20 : null

21 : undef

23 : foreach

24 XX : ldc.8s - load 8-bit signed integer constant XX on stack, where -128 ⇐ XX ⇐ 127

25 XX YY : ldc.15u - load 15-bit unsigned integer constant with value (YY « 7) + (XX & 0x7f) on stack

I.e. XX always has the high bit set, and only delivers 7 bits of information.
Examples:
0x0080 -> 25 80 01   0180
0x0081 -> 25 81 01   0181
0x00FF -> 25 FF 01   01FF
0x0100 -> 25 80 02   0280
0x012c -> 25 ac 02   02ac
0x015e -> 25 de 02   02de
0x017f -> 25 ff 02   02ff
0x0180 -> 25 08 03   0380
0x0190 -> 25 90 03   0390
0x7530 -> 25 b0 ea   eab0
0x7fff -> 25 ff ff   ffff

26 : ldtrue - load Boolean constant true.

27 : ldfalse - load Boolean constant false.

28 : ldnan

29 : pop - remove the top element of the stack.

2A XX : dup ?

2B : catchdone ?

2C XX : ldc.string - load string constant #XX on stack.

2D : ldc.intref

2F : ldc.float

47 : returnvoid - returns from function/method without passing any return value (equal to C/C++ “return;”)

5D XX : ?

62 XX : ?

63 XX : ?

66 XX : load type?

"foo = NaN;" can become
5D 03
66 03
6F 01

6E XX : ldloca.number(?) - load reference to Number variable #XX on stack.

6F XX : ldloca.string - load reference to String variable #XX on stack

80 02 : st - store or assign (.

"foo = 2;" migh be translated to 
6F 01  (push reference to variable foo)
24 02  (push integer constant 2)
80 02  (store)

87 : convert - type conversion? Seen with “as” keyword

90 : neg

91 : inc - increment ++

93 : dec - decrement –

96 : not - Boolean not !

97 : bnot - binary not/bitwise complement ~

A0 : add - addition of numbers and string concatenation +

A1 : sub

A2 : mul

A3 : div - division /

A4 : mod - modulo %

A5 : shl - shift left «

A6 : shr - signed shift right »

A7 : shr.u - unsigned shift right »>

A8 : and - &

A9 : or - |

AA : xor - ^

AB : ceq - compare equal ==

AC : cseq - compare strict equal (no type coercion) ===

AD : clt - compare less than <

AE : cle - compare less than or equal ⇐

AF : cgt - compare greater than >

B0 : cge - compare greater than or equal >=

B3 : is ? is seen with use of “is” keyword

B4 : in

C0 : inc.i

C1 : dec.i

C5 : add.number - addition of numbers +

C6 : sub - subtract -

C7 : mul - multiply *

F0 XX : debugline (XX denotes a line number)